alexa An Integrated Method for Detecting Micro RNA Target Proteins through Reverse-phase Protein Arrays | Open Access Journals
ISSN: 0974-7230
Journal of Computer Science & Systems Biology
Like us on:
Make the best use of Scientific Research and information from our 700+ peer reviewed, Open Access Journals that operates with the help of 50,000+ Editorial Board Members and esteemed reviewers and 1000+ Scientific associations in Medical, Clinical, Pharmaceutical, Engineering, Technology and Management Fields.
Meet Inspiring Speakers and Experts at our 3000+ Global Conferenceseries Events with over 600+ Conferences, 1200+ Symposiums and 1200+ Workshops on
Medical, Pharma, Engineering, Science, Technology and Business

An Integrated Method for Detecting Micro RNA Target Proteins through Reverse-phase Protein Arrays

Jiawen Zhu1, Song Wu1 and Jie Yang2*

1Department of Applied Mathematics and Statistics, Stony Brook University, Stony Brook, NY 11790, USA

2Department of Preventive Medicine, Stony Brook University, Stony Brook, NY 11790, USA

*Corresponding Author:
Jie Yang
Department of Preventive Medicine
Stony Brook University
Stony Brook, NY 11790, USA
Tel: 631-444-2191
Fax: +1 631-444-7525
E-mail: [email protected]

Received date: November 03, 2014; Accepted date: November 18, 2014; Published date: January 01, 2015

Citation: Zhu J, Wu S, Yang J (2015) An Integrated Method for Detecting Micro RNA Target Proteins through Reverse-phase Protein Arrays. J Comput Sci Syst Biol 8:012-033. doi:10.4172/jcsb.1000166

Copyright: ©2015 Zhu J, et al. This is an open-access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.

Visit for more related articles at Journal of Computer Science & Systems Biology

Abstract

Objective: Understanding functions of microRNAs (or miRNAs), particularly their effects on protein degradation, is biologically important. Emerging technologies, including the reverse-phase protein array (RPPA) for quantifying protein concentration and RNA-seq for quantifying miRNA expression, provide a unique opportunity to study miRNA-protein regulatory mechanisms. One naïve way to analyze such data is to directly examine the correlation between the raw miRNA measurements and protein concentrations estimated from RPPA. However, the uncertainty associated with protein concentration estimates is ignored, which may lead to less accurate results and significant power loss.

Methods: We propose an integrated nonlinear hierarchical model for detecting miRNA targets through original RPPA intensity data. This model is fitted within a maximum likelihood framework and the correlation test between miRNA and protein is assessed using Wald tests. We compare this model and the simple method through extensive simulation studies and a real dataset from the Cancer Genome Atlas (TCGA) project.

Results: This integrated method is shown to have consistently higher power than the simple method, especially when sample sizes are limited and when the RPPA intensity levels are close to the boundaries of imaging limits.

Conclusions: Our proposed method is powerful in detecting miRNA’s protein target through RPPA. We recommend this method in practice.

Keywords

MicroRNA; Reverse-phase Protein Arrays; Nonlinear mixed model; Micro RNA target

Introduction

MicroRNA (miRNA) is a set of small, non-coding RNA molecules that can post-transcriptionally regulate a broad range of gene expression in both plants and animals. They have been suggested to be involved in many important biological processes, such as normal physiological development and disease onsets. In the past decade, many efforts have been put to search for the miRNA targets [1]. Although our understanding on some miRNAs has been dramatically improved, as of today, the targets of many others remain largely unknown. Therefore, powerful methods for efficient detection of miRNA targets are still in great need.

In general, miRNA regulates the expression of its target genes through two mechanisms–mRNA degradation or translation inhibition. That is, if a miRNA and its target gene can complement extensively, the miRNA-mRNA target may form a double-strand RNA (dsRNA) structure, after which, the mRNA can be cleaved and degraded to lower the mRNA expression and subsequently protein expression [2,3]. On the other hand, if a miRNA and its target can only complement partially, the target mRNA will not be directly degraded but its translation may be repressed [4,5]. So, in both mechanisms, the total protein level relating to the miRNA targets would be reduced, resulting in their functional losses.

Based on the phenomenon that the sequences of miRNAs and their target genes complement to each other, or at least partially, one way of the miRNA target identification is through in silico prediction. Several software tools have been developed for such purpose, each with its own unique feature. For example, miRanda scored the likelihoods of mRNA down-regulation according to a regression model that is trained on sequence and contextual features of the predicted miRNA::mRNA duplex [6], and Target Scan studies on the miRNA::mRNA duplex interactions according to a thermodynamics-based modeling and comparative sequence analysis [7]. Based on these computational tools, several databases with predicted miRNA targets have been generated (microRNA.org and targetscan.org). However, one major limitation is that they all suffer from large percentage of false positives, which hinder their practical usage.

Another popular way to determine the miRNA targets is through experimental data by measuring downstream effects of miRNAs. Since miRNAs can induce protein reduction via both functional mechanisms, protein expression seems to be the right mark. However, due to difficulties in high- throughput quantification of protein expression but relative ease in that of mRNA, conventionally, scanning of the miRNA targets is mainly through testing negative correlations between miRNAs and mRNAs. For example, high-throughput techniques, such as miRNA and mRNA gene microarray, can be applied to measure their expression levels, and then the correlations analyses can be conducted subsequently to filter out miRNA-mRNA pairs that show significant negative correlations as potential candidates for further analyses [8]. More recently, with the advent and rapid advance of sequencing techniques, the miRNA sequencing (miRNA-seq) and RNA sequencing (RNA-seq) platforms have become more and more popular for the quantification of the miRNA and RNA expressions. The main drawback for the miRNA/mRNA correlation analyses is that they can only identify miRNA targets that may change at mRNA levels, but is determined to fail for those modulated through translation inhibition. Therefore, miRNA/mRNA correlation analysis is only able to find partial targets of miRNAs.

It is only until recently that a new technology, called the reverse phase protein array (RPPA) or protein lysate array has been developed to simultaneously quantify the protein expression in a large sample cohort [9]. In contrast to the usual array design that quantifies expression levels of multiple genes in one sample, the RPPA measures the protein expression levels of many samples on one array. Usually, a 2-fold serial dilution of samples is used to avoid signal saturation for very high protein concentration. Since the RPPA data are measured through image signals, which are characterized by background noises at the low end and saturation signals at the high end, a sigmoid-like response curve model has been used to estimate protein concentration [10,11]. Additionally, some flexible nonparametric joint sample models have also been proposed [12]. In principle, we can directly use the protein concentrations estimated from these methods to correlate with miRNA expression for miRNA target identification. However, in this case, the uncertainties associated with the protein concentration estimates are ignored, leading to less accurate and less powerful inference.

In this article, we propose an integrated hierarchical model to detect miRNA targets based on protein expression data measured by the recently emerged RPPA and high throughput miRNA- seq data. Extensive Monte Carlo simulations have been performed to examine the performance of our proposed model under different scenarios. This model was further applied to a real dataset from TCGA to demonstrate its practical use.

Methods

PPA

To quantify the protein expression on a RPPA array, a sigmoidal model is commonly assumed to describe the relationship between the intensity level and the protein concentration with additive error [10,13]:

equation (1)

wher yij is the gray-level intensity from sample at ith diilution, i=1,…I and j=1,…,j, Xi is the binary logarithm of the median effective protein concentration level (EC50), a single quantity per dilution series to represent the concentration of the protein,xi+lj is the binary logarithm of the protein concentration after jth dilution whereequation εi j is the error term assumed to have a normal distribution with mean 0 and equation,and variance β={β123} 1Since equationand equation1 is interpreted as the lowest intensity level without noise, and β2 is the increment from the lowest to the highest intensity or the saturation level.

To estimate the relative protein concentration in RPPA, one algorithm based on logistic model fitting used by Hu [12] is given as follows:

The initial intensity data are first transformed as

equation (2)

and initial β are set as

equation

range(y)=max(Y)−min (Y).

The initial median effective protein concentration level Xi are estimated by using:

equation (3)

where y linear , is the mean value among { y linear}

To update the parameters equation in the nonlinear model, the nonlinear least-squares estimates of equationwere calculated based on the following model [14]:

equation (4)

After obtaining equationthe nonlinear least-square method is used again to update the relative protein level X={xi,i=1…I} andequation.This iteration continues until convergence.

A naïve model for correlating miRNA and protein expression

Since X={xii=1,2,…I}, the log2 transformation of the median effective protein concentration levels, can be estimated from the RPPA, a straightforward way to examine the relationship between miRNA and protein expression levels is through Pearson’s correlation coefficients or simple linear regression models, which is referred as the naïve model in this article.

The linear relationship between protein and miRNA in the naïve model can be expressed as equation , whereequationand {Zi=1,2…} are log-transformed expression levels of a specific miRNA from sample i=1,2…I.

The parameter estimates equationcan be calculated by using linear regression. equationis our parameter of interest, describing the correlation between a miRNA and protein pair. We can test H0: α2=0 to determine if a particular pair of miRNA and protein is related or not.

A hierarchical model for correlating miRNA and protein expression

Although the naïve method is straightforward, uncertainty associated with protein concentration estimates is ignored. As demonstrated in later sections, it leads to less accurate results and significant power loss. Here we propose a nonlinear hierarchical model for studying the relationship between miRNA and protein expression, in which the correlation analysis is integrated with the estimation of protein concentration. The model is given as follows:

equation (5)

equation

equation (6)

Here (.) is a general function to describe how xi, the protein level, and Zi, the miRNA expression level, is related. β= {β1,23} is the parameter vector of the response curve function g(.). To directly compare with the naïve model, we assume (Zi) to be linear in equation (6), that is,equationWe further assume that the two error terms,ηi, and εij are independent of each other. In this hierarchical framework, the relationship between miRNA and protein expression level can be estimated without explicitly quantifying the protein concentration based on intensity data first. This model is referred as the integrated model hereinafter.

The likelihood function for Yand Zcan be written as a joint probability function

equation (7)

where Ø={α12123} is a vector including parameters in function f(.) and g(.), Y={yij;| i=1,2…I,j=1,2…j} representing the RPPA intensity levels and Z={Zi|i=1,2…I} representing the log-transformed miRNA expression levels.

Computational algorithm

The unknown parameters Ø={α12123} can be estimated within the maximum likelihood framework. The adaptive Gaussian quadrature method is used to approximate the integral and the dual quasi-Newton method can be further applied in maximizing the likelihood function [15-17]. Our computational algorithm is illustrated in Figure 1. This algorithm is implemented in SAS nlmixed procedure and a SAS macro is available upon request to fit the integrated model.

computer-science-systems-biology-Flow-chart

Figure 1: Flow chart of computational algorithm to fit the integrated model based on adaptive Gaussian quadrature method and dual quasi-Newton algorithm. Step 0: estimate the initial value of ��,0,��1, denoted as ��(0), by using the naïve model; Step 1: generate the approximate likelihood function by using adaptive Gaussian Quadrature method; Step 2: compute the quasi-Newton direction Δ��, determine the step size �� to satisfy the Goldstein conditions; Step 3: update parameters �� value; Step 4: update Hessian matrix; Step 5: check if the iteration stops. If not, go to Step 2, if yes, the iteration stops. A grid searching with center from the estimates of naïve model was applied in our algorithm.

Based on the idea of Hermite integration for function (x)

equation (8)

Our likelihood function can be written as

equation

equation (9)

m is the number of quadrature points which is set to 5 in our analysis.xkand wk denote the standard Gauss-Hermite abscissas and weights ηi minimizes

equation (10)

and Γi is the Hessian matrix from the minimization

With the approximate likelihood function, we employed the quasi- Newton algorithm for parameter estimation. Unlike the Newton algorithm, approximate Hessian matrix was used instead of true Hessian matrix in the quasi-Newton algorithm. The step size t was calculated by quadratic interpolation and cubic extrapolation and BFGS method was used to update the approximate Hessian matrix. A grid searching with center from the estimates of Naïve model was applied in our algorithm.

Hypothesis testing

Since it is expected that miRNA negatively regulate the protein level of its target gene, to test if there is a significant relationship between a specific pair of miRNA and protein, the hypothesis test can set up as a one-sided test:

H0: α2=0 vs H a2<0

Once the maximum likelihood estimates are obtained, a likelihood ratio test (LRT), a Wald test or a Score test can be constructed. However, LRT can be very time consuming and is not appropriate for one–sided test. For Score test, the confidence interval ofα2 is difficult to be calculated. Thus Wald tests were used in our simulation and real data example, and its test statistic is:

equationunder H0, where (Ø) represent the fisher

information matrix of the likelihood function. The null hypothesis was rejected when p-value was above 0.05 in our simulation study.

Simulation studies

Extensive simulation studies were carried out to examine the performance of our proposed integrated model and to compare with the naïve model approach. Protein intensities were generated by using a sigmoidal response curve (Figure 2a). And a typical miRNA expression distribution in the TCGA data was borrowed in this simulation to mimic the real data and generate protein EC50S (Figure 2b). Also, the true values of {β1,2301} were set as {50, 30000, 1, 1, and 300} to mimic parameter values estimated from a real TCGA ovarian cancer data set. Different strengths of correlation between miRNA and protein expression levels, as characterized by α2, were examined in a range from 0, which represents the null hypothesis, to -1.5, which yields the power of 1 for the integrated method. In order to investigate the performance of two models with protein intensity values located in different areas of the sigmoidal curve, α1 was set as 0 and 5 corresponding to the middle part and upper part of sigmoidal curve, respectively. The upper part of a sigmoidal curve corresponds to a scenario where most of intensity levels are close to the saturation point. The RPPA intensity levels range between 10 and 30100. An illustration of the sigmoidal curve used to generate simulated data was showed in Figure 2a. The locations of protein intensity center were marked by circles. If simulated intensity values are beyond the imaging boundary, they would be replaced with the boundary value with small error (Gaussian distributed with mean 0 and standard deviation 5). 1000 simulations were carried out for each parameter setting under different sample sizes (N=20, 50, 100 and 300). Generally, there are 5 diluted samples in one dilution series, so j=5 were used in our simulation setting. Pre-specified Type I error was set to be 0.05.

computer-science-systems-biology-sigmoidal-shape

Figure 2: An illustration of (a) a sigmoidal shape response curve. When ��1 was set to be 0, the center of the EC50s would located at 0; When ��1 was set to be 5, the center of EC50s would located at 5; (b) distribution of a typical miRNA expression in the TCGA data we used in the simulation.

The false positive rates and detection powers for miRNA targets for both the integrated model and the naïve model under different sample sizes were shown in Figures 3 and 4. It is clear that when there was no relationship between miRNA and protein (α2=0), both models can well control the pre-specified type-I error when sample size were bigger than 50. The integrated model was consistently more powerful than the naïve model, especially when the RPPA intensity levels are close to the boundaries of imaging limits (Figure 4). Table A1-A2 in appendix listed the detailed point estimates of all parameters and their corresponding standard errors. Figure A1 and A2 in appendix illustrated the variations of the point estimates of α2 under different simulation settings. The integrated model consistently yielded parameter estimates of α2 with similar or much less standard errors than those from the naïve model. When the RPPA intensities reached the upper flatter part of the sigmoidal curve, which caused information loss because of intensity level truncation at the saturation points, both the naïve and the integrated method over-estimated β1, which represents the lower imaging limits. However, in this situation the integrated method still

computer-science-systems-biology-naive-model

Figure 3: Power curve of the naïve model (solid line) and the integrated model (dashed line) according to different simulation scenarios: sample size ranged from 20 to 300 and the protein intensities were located in the middle part of a sigmoidal curve. Detection powers (type I error if ��2=0) denoted by ��1 and ��2 under different correlation strengths were report on the bottom of each plot for the naïve and integrated models, respectively. Both models can well control the pre-specified type-I error when sample size were bigger than 50. Two models had similar detection performance, especially when sample size increased.

computer-science-systems-biology-Power-curves

Figure 4: Power curves of the naïve model (solid line) and the integrated model (dashed line) according different simulation scenarios: sample size ranged from 20 to 300 and the protein intensities were located in the upper part of a sigmoidal curve. Detection powers (type I error if ��2=0) denoted by ��1 and ��2 under different correlation strength were report on the bottom of each figure for the naïve and integrated models, respectively. Both models can well control the pre-specified type-I error when sample size were bigger than 50. The integrated model was consistently more powerful than the naïve model.

Analysis of TCGA Ovarian Cancer Data

Both models were applied onto an ovarian cancer dataset from the TCGA project. In this dataset, there were 333 ovarian cancer samples with both miRNA and RPPA data available. 352 miRNAs having more than 50% of non-zero counts and 165 proteins were included in our analyses.

The results from both naïve and integrated models on predicting miRNA targets were reported in Table 1. False Discover Rate (FDR) at 10% was used to adjust for multiple testing [18]. The integrated model approach we proposed found 1106 potential miRNA-protein pairs, 797 of which were on non-phosphorylated protein array. 822 pairs were found on non-phosphorylated protein array: 250 out of them were found by integrated model only and 25 pairs were found by naïve model only. Integrated model found significantly more number of potential miRNA-protein pairs (P<0.0001 according to McNemar’s test). Furthermore, we compared our results with miRNA targets identified by miRanda algorithm [19-22]. 98 targets, which were found by both the integrated and the naïve model, and 31 targets, which were found only by the integrated model, were confirmed by miRanda database. However, only 6 targets found by the naïve model only were confirmed by miRanda database. MirTarbase, a dataset based on manually surveying pertinent literature [23] was used to further verify our results. 15 suggested targets found by both the integrated and the naïve model were supported by the MirTarBase dataset. 11 of the 15 suggested targets found by both the integrated and the naïve model were supported by strong experimental evidences according to the MirTarBase dataset. One suggested target found by the integrated model only were supported by strong experimental evidences according to the MirTarBase dataset. None of the suggested targets found by the naïve model only were supported by strong experimental evidences according to the MirTarBase dataset. This suggests that there could be a number of undiscovered miRNA targets included in the findings of integrated and naïve models. The list of 822 miRNA/protein pairs was included in the appendix (Table A3).

  α1 α2 β1 β2 β2 σ0 σ1
Sample size 20            
true value 0 0 50 30000 1 1 500
naive model 0.0213 (0.0517) -0.0063 (0.0117) 78.13 (22.56) 29958.77 (31.48) 1.0022 (0.0014) 0.9809 (0.0053) 463.61 (1.34)
integrated model 0.015 (0.052) -0.0062 (0.0117) 142.07 (15.3) 29881.73 (28.8) 1.0077 (0.0014) 0.9298 (0.0051) 490.89 (1.24)
true value 0 -0.1 50 30000 1 1 500
naive model -0.0062 (0.0057) -0.1066 (0.0117) 78.43 (22.69) 29954.12 (31.17) 1.0023 (0.0014) 0.981 (0.0053) 463.75 (1.34)
integrated model -0.0106 (0.0068) -0.1067 (0.0117) 136.82 (15.25) 29875.35 (28.62) 1.0078 (0.0014) 0.9299 (0.005) 490.51 (1.23)
true value 0 -0.3 50 30000 1 1 500
naive model -0.0015 (0.0057) -0.3067 (0.0117) 83.04 (23.4) 29904.9 (31.25) 1.0044 (0.0014) 0.9811 (0.0053) 464.85 (1.38)
integrated model -0.0126 (0.0068) -0.3071 (0.0117) 168.89 (15.35) 29827.69 (29.14) 1.0102 (0.0014) 0.93 (0.0051) 491.49 (1.23)
true value 0 -0.5 50 30000 1 1 500
naive model -0.0015 (0.0057) -0.5063 (0.0117) 81.23 (22.74) 29909.99 (30.21) 1.0045 (0.0014) 0.9807 (0.0053) 464.58 (1.36)
integrated model -0.0119 (0.0068) -0.5069 (0.0117) 161.52 (14.84) 29838.19 (28.02) 1.0096 (0.0014) 0.9297 (0.0051) 491.18 (1.23)
true value 0 -1 50 30000 1 1 500
naive model -0.0043 (0.0061) -1.0021 (0.0117) 23.03 (20.93) 30048.43 (26.41) 0.9997 (0.0012) 0.9795 (0.0053) 465.42 (1.4)
integrated model -0.0105 (0.0068) -1.005 (0.0117) 91.21 (13.22) 29964.21 (24.28) 1.003 (0.0012) 0.9298 (0.0051) 491.06 (1.27)
true value 0 -1.3 50 30000 1 1 500
naive model -0.0055 (0.006) -1.3033 (0.0117) 39.67 (18.69) 30027.79 (24.25) 1.0018 (0.0012) 0.9792 (0.0053) 465.7 (1.37)
integrated model -0.0099 (0.0067) -1.3077 (0.0118) 96.34 (11.97) 29952.63 (22.46) 1.0035 (0.0011) 0.9301 (0.005) 491.25 (1.25)
true value 0 -1.5 50 30000 1 1 500
naive model -0.0096 (0.0063) -1.5032 (0.0117) 83.81 (17.38) 29973.69 (22.32) 1.0042 (0.0011) 0.9789 (0.0053) 464.47 (1.35)
integrated model -0.0111 (0.0069) -1.5086 (0.0117) 124.24 (11.18) 29909.02 (20.75) 1.0054 (0.0011) 0.93 (0.0051) 491.11 (1.28)
Sample size 50            
true value 0 0 50 30000 1 1 500
naive model -0.023 (0.0283) 0.0054 (0.0064) -11.3 (16.58) 30137.24 (20.17) 0.9952 (9e-04) 0.9923 (0.0032) 457.55 (0.85)
integrated model -0.0259 (0.0284) 0.0053 (0.0064) 43.36 (9.77) 30057.24 (18.46) 0.9982 (9e-04) 0.9719 (0.0032) 495.97 (0.79)
true value 0 -0.1 50 30000 1 1 500
naive model -0.0012 (0.0041) -0.0946 (0.0064) -6.46 (16.12) 30141.23 (19.69) 0.9951 (9e-04) 0.9925 (0.0032) 457.14 (0.84)
integrated model -0.0026 (0.0043) -0.0947 (0.0064) 39.43 (9.46) 30061.21 (18.05) 0.9982 (8e-04) 0.972 (0.0032) 495.79 (0.79)
true value 0 -0.3 50 30000 1 1 500
naive model 0.0021 (0.0041) -0.2942 (0.0064) -14.38 (15.88) 30131.06 (18.59) 0.9954 (8e-04) 0.9923 (0.0032) 458.04 (0.86)
integrated model -9e-04 (0.0044) -0.2945 (0.0064) 36.9 (8.99) 30053.19 (17.01) 0.9983 (8e-04) 0.9719 (0.0032) 496.54 (0.78)
true value 0 -0.5 50 30000 1 1 500
naive model 0.0031 (0.004) -0.4939 (0.0064) -10.89 (15.49) 30113.24 (17.77) 0.9965 (8e-04) 0.9919 (0.0032) 457 (0.84)
integrated model -0.002 (0.0044) -0.4946 (0.0064) 46.42 (8.56) 30042.02 (16.43) 0.9988 (8e-04) 0.9716 (0.0032) 496.05 (0.79)
true value 0 -1 50 30000 1 1 500
naive model -0.0031 (0.0042) -0.9927 (0.0064) 16.45 (13.86) 30112.04 (16.06) 0.998 (8e-04) 0.9915 (0.0032) 457.22 (0.82)
integrated model -0.0022 (0.0044) -0.9948 (0.0064) 46.61 (7.86) 30043.54 (14.75) 0.9986 (7e-04) 0.9719 (0.0032) 496.19 (0.79)
true value 0 -1.3 50 30000 1 1 500
naive model -0.0068 (0.0043) -1.2911 (0.0064) 48.01 (13.45) 30080.17 (14.29) 1.0002 (7e-04) 0.9911 (0.0032) 457.21 (0.82)
integrated model -0.0018 (0.0044) -1.2949 (0.0064) 60.83 (7.18) 30013.75 (13.47) 0.9997 (7e-04) 0.9721 (0.0032) 496.15 (0.79)
true value 0 -1.5 50 30000 1 1 500
naive model -0.0078 (0.0041) -1.4898 (0.0064) 77.09 (11.46) 30027.67 (13.43) 1.0032 (7e-04) 0.9909 (0.0032) 456.69 (0.78)
integrated model -0.0026 (0.0043) -1.4944 (0.0064) 82.21 (6.6) 29975.26 (12.74) 1.0017 (7e-04) 0.972 (0.0032) 496.5 (0.77)
Sample size 100            
true value 0 0 50 30000 1 1 500
naive model 0.0124 (0.0207) -0.0032 (0.0047) -21.78 (12.82) 30189.17 (14.32) 0.9927 (6e-04) 0.997 (0.0022) 455.17 (0.62)
integrated model 0.0143 (0.0208) -0.0033 (0.0047) 7.17 (6.74) 30116.73 (12.96) 0.9951 (6e-04) 0.9863 (0.0022) 498.45 (0.57)
true value 0 -0.1 50 30000 1 1 500
naive model 0.0029 (0.0032) -0.1028 (0.0047) -46.45 (12.49) 30192.3 (13.75) 0.9929 (6e-04) 0.9965 (0.0022) 454.27 (0.6)
integrated model -5e-04 (0.0032) -0.1029 (0.0047) 1.55 (6.47) 30125.41 (12.4) 0.9948 (6e-04) 0.9859 (0.0022) 498.08 (0.56)
true value 0 -0.3 50 30000 1 1 500
naive model 3e-04 (0.0033) -0.3033 (0.0047) -26.94 (12.46) 30177.98 (13.77) 0.9932 (6e-04) 0.9968 (0.0022) 455.26 (0.6)
integrated model -9e-04 (0.0032) -0.3036 (0.0047) 11.3 (6.47) 30110.92 (12.44) 0.9951 (6e-04) 0.9861 (0.0022) 498.77 (0.56)
true value 0 -0.5 50 30000 1 1 500
naive model 0.0034 (0.0033) -0.5024 (0.0047) -29.46 (12.41) 30152.94 (12.87) 0.9948 (6e-04) 0.9968 (0.0022) 454.68 (0.59)
integrated model -0.002 (0.0032) -0.5029 (0.0047) 23.79 (6.41) 30091.34 (11.81) 0.996 (6e-04) 0.9863 (0.0022) 498.68 (0.56)
true value 0 -1 50 30000 1 1 500
naive model -0.0063 (0.0033) -1.0002 (0.0047) 18.58 (10.56) 30140.4 (11.11) 0.9961 (5e-04) 0.9965 (0.0022) 454.05 (0.58)
integrated model -0.0024 (0.0032) -1.0018 (0.0047) 31.28 (5.38) 30081.86 (10.35) 0.9964 (5e-04) 0.9864 (0.0022) 498.21 (0.57)
true value 0 -1.3 50 30000 1 1 500
naive model -0.0072 (0.0033) -1.3008 (0.0047) 47.51 (9.63) 30088.26 (9.9) 0.9992 (5e-04) 0.9967 (0.0022) 453.78 (0.59)
integrated model -0.0024 (0.0032) -1.3036 (0.0047) 52.33 (4.87) 30038.92 (9.46) 0.9982 (5e-04) 0.9871 (0.0022) 498.32 (0.59)
true value 0 -1.5 50 30000 1 1 500
naive model -0.0066 (0.0032) -1.5001 (0.0047) 65.64 (8.32) 30046.58 (8.83) 1.0018 (4e-04) 0.996 (0.0022) 452.58 (0.56)
integrated model -8e-04 (0.0032) -1.5037 (0.0047) 62.17 (4.51) 30006.08 (8.55) 0.9998 (4e-04) 0.9864 (0.0022) 498.25 (0.58)
Sample size 300            
true value 0 0 50 30000 1 1 500
naive model -0.0029 (0.0118) -7e-04 (0.0027) -7.69 (8.36) 30141.18 (8.44) 0.9945 (4e-04) 0.9998 (0.0014) 451.18 (0.34)
integrated model -0.0039 (0.0118) -7e-04 (0.0027) 17.74 (3.87) 30099.42 (7.63) 0.9956 (4e-04) 0.9953 (0.0014) 499.45 (0.32)
true value 0 -0.1 50 30000 1 1 500
naive model 3e-04 (0.0023) -0.1009 (0.0026) -45.21 (8.74) 30156.77 (7.84) 0.9938 (4e-04) 0.9999 (0.0014) 451.88 (0.34)
integrated model -0.0084 (0.002) -0.1009 (0.0026) 11.41 (3.65) 30117.72 (7.15) 0.9947 (3e-04) 0.9955 (0.0014) 500 (0.32)
true value 0 -0.3 50 30000 1 1 500
naive model -0.0038 (0.0024) -0.3005 (0.0027) -19.8 (8.81) 30144.6 (8.1) 0.9945 (4e-04) 0.9996 (0.0014) 451.86 (0.35)
integrated model -0.0069 (0.0019) -0.3006 (0.0027) 14.95 (3.71) 30101.39 (7.38) 0.9955 (4e-04) 0.9953 (0.0014) 499.85 (0.31)
true value 0 -0.5 50 30000 1 1 500
naive model -0.0041 (0.0023) -0.5006 (0.0026) -16.25 (8.32) 30142.07 (7.35) 0.9948 (3e-04) 0.9996 (0.0014) 451.4 (0.33)
integrated model -0.0085 (0.0019) -0.5009 (0.0026) 22.01 (3.48) 30102.75 (6.82) 0.9954 (3e-04) 0.9953 (0.0014) 499.8 (0.32)
true value 0 -1 50 30000 1 1 500
naive model -0.0121 (0.0023) -0.999 (0.0027) 30.28 (7.06) 30107.97 (6.46) 0.9968 (3e-04) 0.9996 (0.0014) 451 (0.33)
integrated model -0.0082 (0.002) -0.9999 (0.0027) 32.49 (3.08) 30070.72 (6.22) 0.9967 (3e-04) 0.9955 (0.0014) 499.92 (0.33)
true value 0 -1.3 50 30000 1 1 500
naive model -0.0101 (0.0021) -1.2999 (0.0028) 47.45 (5.33) 30058.55 (5.39) 0.9997 (3e-04) 1.0019 (0.0027) 448.96 (0.31)
integrated model -0.0077 (0.002) -1.3008 (0.0028) 49.28 (2.79) 30036.58 (5.22) 0.9983 (3e-04) 0.9955 (0.0014) 499.31 (0.34)
true value 0 -1.5 50 30000 1 1 500
naive model -0.0106 (0.0021) -1.4962 (0.0027) 47.23 (4.56) 30059.42 (5.23) 1.0004 (3e-04) 0.9997 (0.0015) 450.03 (1.52)
integrated model -0.0091 (0.0021) -1.4986 (0.0028) 49.02 (2.61) 30043.6 (5.15) 0.9979 (3e-04) 0.9955 (0.0014) 499.52 (0.35)

Table 1: Table for the detailed point estimates of all unknown parameters and their standard error;Sample size was from 20 to 300 and the protein intensity located in the middle part of sigmoidal curve; the integrated model had a similar performance as the naïve model.

  α1 α2 β1 β2 β3 σ0 σ1
Sample size 20            
true value 5 0 50 30000 1 1 500
naive model 2.3538 (0.0337) 0.0108 (0.0348) 21673.99 (73.67) 8063.19 (72.22) 1.8432 (0.0062) 1.5515 (0.1377) 505.9 (4.52)
integrated model 4.4653 (0.08) -0.0161 (0.0139) -274276.76 (52197.04) 304133.74 (52200.42) 1.2805 (0.011) 0.9099 (0.0111) 457.39 (1.95)
true value 5 -0.1 50 30000 1 1 500
naive model 2.3656 (0.0399) -0.0895 (0.0592) 21700.84 (71.7) 8041.22 (70.25) 1.8401 (0.0059) 1.6134 (0.1588) 504.84 (3.24)
integrated model 4.5686 (0.0859) -0.0693 (0.0418) -330800.61 (56594.51) 360639.78 (56591.54) 1.2918 (0.0288) 0.9732 (0.0742) 458.96 (2)
true value 5 -0.3 50 30000 1 1 500
naive model 2.3695 (0.0275) -0.3577 (0.0482) 21516.03 (74.6) 8221.13 (73.15) 1.8385 (0.0057) 1.5135 (0.1068) 505.61 (3.97)
integrated model 4.403 (0.0769) -0.308 (0.0123) -249677.33 (50408.57) 279512.95 (50405.41) 1.2762 (0.0105) 0.8937 (0.0099) 456.25 (1.72)
true value 5 -0.5 50 30000 1 1 500
naive model 2.4446 (0.0374) -0.7107 (0.0782) 21269.05 (74.86) 8464.35 (73.51) 1.8481 (0.0059) 1.6564 (0.1521) 507.56 (2.77)
integrated model 4.5227 (0.0783) -0.5035 (0.0123) -250395.86 (48590.31) 280258.79 (48592.42) 1.2486 (0.0097) 0.8891 (0.0102) 457.35 (1.82)
true value 5 -1 50 30000 1 1 500
naive model 2.7712 (0.0505) -1.4615 (0.1208) 19902.57 (90.02) 9814.9 (88.76) 1.8388 (0.0058) 2.1685 (0.1959) 527.34 (4.87)
integrated model 4.1187 (0.0636) -1.0024 (0.0148) -52274.05 (21283.7) 82100.65 (21281.33) 1.2984 (0.0263) 0.8417 (0.0365) 469.74 (2.81)
true value 5 -1.3 50 30000 1 1 500
naive model 2.9411 (0.0548) -1.7328 (0.0838) 18770.02 (98.58) 10936.33 (97.49) 1.828 (0.0054) 2.2838 (0.2277) 539.5 (4.31)
integrated model 4.1799 (0.0528) -1.3159 (0.0145) -81574.65 (30566.75) 111394.19 (30567.8) 1.2845 (0.0095) 0.8634 (0.0125) 478.65 (2.39)
true value 5 -1.5 50 30000 1 1 500
naive model 3.2206 (0.0717) -2.3778 (0.1593) 18002.82 (105.51) 11696.6 (104.59) 1.8231 (0.0057) 2.9557 (0.2621) 560.41 (7.25)
integrated model 3.6493 (0.6203) -0.964 (0.6069) -51974.4 (22391.12) 81776.04 (22386.56) 1.2932 (0.0201) 0.8548 (0.0355) 488.64 (3.71)
Sample size 50            
true value 5 0 50 30000 1 1 500
naive model 2.7437 (0.0369) -0.0469 (0.0526) 20499.39 (63.98) 9225.41 (62.87) 1.7402 (0.0037) 2.7635 (0.2377) 485.25 (2.14)
integrated model 3.5198 (0.0405) 0.0089 (0.0067) 8598.57 (1581.12) 21172.62 (1582.55) 1.4285 (0.0103) 0.8595 (0.0138) 489.77 (1.59)
true value 5 -0.1 50 30000 1 1 500
naive model 2.7316 (0.0347) -0.1643 (0.0384) 20463.66 (65.62) 9261.85 (64.57) 1.738 (0.0038) 2.6338 (0.2266) 484.29 (1.72)
integrated model 3.4811 (0.0385) -0.0898 (0.0067) 9927.86 (1621.29) 19842.78 (1622.67) 1.434 (0.0104) 0.8728 (0.0129) 491.19 (1.61)
true value 5 -0.3 50 30000 1 1 500
naive model 2.8069 (0.0401) -0.4472 (0.047) 20293.49 (66.02) 9426.7 (65.05) 1.7416 (0.0037) 2.9565 (0.2576) 486.58 (1.97)
integrated model 3.4894 (0.0357) -0.2882 (0.0067) 12135.89 (534.51) 17635.35 (537.21) 1.4246 (0.0101) 0.8701 (0.0131) 491.18 (1.66)
true value 5 -0.5 50 30000 1 1 500
naive model 2.828 (0.0387) -0.6441 (0.0451) 19946.91 (69.44) 9770.55 (68.46) 1.7402 (0.0038) 2.8163 (0.2509) 491.23 (2.77)
integrated model 3.5327 (0.0365) -0.4769 (0.0067) 11365.82 (735.99) 18404.38 (738.2) 1.4188 (0.01) 0.8762 (0.0124) 493.4 (1.7)
true value 5 -1 50 30000 1 1 500
naive model 3.2183 (0.047) -1.7113 (0.0905) 18367.63 (79.4) 11331.51 (78.56) 1.7405 (0.0037) 3.9333 (0.301) 506.65 (2.69)
integrated model 3.4259 (0.029) -0.9716 (0.0072) 13477.54 (288.17) 16260.6 (291.47) 1.4497 (0.0086) 0.7908 (0.0171) 512.56 (1.84)
true value 5 -1.3 50 30000 1 1 500
naive model 3.5889 (0.0593) -2.5651 (0.1361) 17174.96 (86.16) 12514.49 (85.49) 1.7375 (0.0037) 5.4373 (0.3774) 521.52 (3.13)
integrated model 3.2866 (0.1729) -1.4043 (0.1428) 13337.94 (260.58) 16383.67 (263.67) 1.4403 (0.0146) 0.8708 (0.1229) 528.37 (2.55)
true value 5 -1.5 50 30000 1 1 500
naive model 3.8296 (0.0621) -2.9831 (0.1416) 16141.61 (89.86) 13546.24 (89.49) 1.7198 (0.0039) 6.1208 (0.3784) 523.94 (2.07)
integrated model 3.305 (0.1669) -1.3648 (0.105) 13436.45 (198.21) 16273.32 (201.3) 1.4225 (0.0332) 0.9029 (0.1719) 536.53 (2.56)
Sample size 100            
true value 5 0 50 30000 1 1 500
naive model 2.9087 (0.0269) -0.039 (0.0434) 19661.6 (56.83) 10058.46 (56.2) 1.6748 (0.003) 3.337 (0.2366) 468.53 (1.22)
integrated model 3.0493 (0.0261) -0.0055 (0.006) 16387.53 (641.05) 13333.07 (642.19) 1.5292 (0.0075) 0.8232 (0.0178) 501.83 (1.1)
true value 5 -0.1 50 30000 1 1 500
naive model 2.9097 (0.0291) -0.1508 (0.043) 19584.67 (57.9) 10137.22 (57.29) 1.6675 (0.0028) 3.2625 (0.2545) 469.76 (1.53)
integrated model 3.0247 (0.0234) -0.0985 (0.0065) 17062.36 (191.54) 12657.31 (194.2) 1.5329 (0.0073) 0.8748 (0.0133) 502.86 (1.06)
true value 5 -0.3 50 30000 1 1 500
naive model 2.9106 (0.0275) -0.3621 (0.0361) 19313.04 (62.78) 10404.89 (62.09) 1.6712 (0.003) 3.0649 (0.2302) 469.78 (0.91)
integrated model 3.0442 (0.0224) -0.3021 (0.0073) 16981.12 (176.62) 12738.09 (179.23) 1.5236 (0.007) 0.8185 (0.0172) 504.55 (1.09)
true value 5 -0.5 50 30000 1 1 500
naive model 3.0186 (0.0323) -0.7679 (0.044) 18919.14 (64.29) 10793.17 (63.66) 1.674 (0.003) 3.701 (0.2824) 474.8 (1.61)
integrated model 3.0784 (0.0226) -0.4853 (0.0049) 16620.4 (198.41) 13095.57 (200.87) 1.5193 (0.0068) 0.8203 (0.0165) 508.61 (1.12)
true value 5 -1 50 30000 1 1 500
naive model 3.5281 (0.0442) -2.0565 (0.0999) 17213.8 (74.03) 12482.15 (73.63) 1.6745 (0.003) 5.9013 (0.3507) 486.43 (0.92)
integrated model 3.1546 (0.0163) -0.9644 (0.0055) 16182.07 (131.53) 13518.68 (133.43) 1.5056 (0.0053) 0.7447 (0.0202) 521.9 (1.12)
true value 5 -1.3 50 30000 1 1 500
naive model 3.8155 (0.0476) -2.6141 (0.0999) 15813.29 (80.09) 13875.57 (79.85) 1.6668 (0.0034) 7.1509 (0.3873) 504.16 (2.87)
integrated model 3.28 (0.0142) -1.2628 (0.0074) 15230.07 (110.54) 14467.41 (111.99) 1.4792 (0.0047) 0.7278 (0.0212) 531.28 (1.16)
true value 5 -1.5 50 30000 1 1 500
naive model 4.23 (0.0626) -3.6045 (0.1408) 14844.17 (83.8) 14844.17 (83.59) 1.6571 (0.0034) 9.335 (0.4778) 513.6 (2.09)
integrated model 3.4028 (0.016) -1.4779 (0.0122) 14282.26 (116.92) 15415.65 (118.47) 1.4536 (0.0046) 0.6573 (0.0252) 539.12 (1.27)
Sample size 300            
true value 5 0 50 30000 1 1 500
naive model 3.0135 (0.0208) 0.0285 (0.0232) 18437.53 (67.92) 11288.37 (67.65) 1.5908 (0.003) 3.2476 (0.263) 451.57 (1.21)
integrated model 2.9 (0.0112) 5e-04 (0.0039) 18348.42 (75.58) 11365.95 (76.43) 1.5243 (0.0043) 0.8799 (0.0187) 498.43 (0.68)
true value 5 -0.1 50 30000 1 1 500
naive model 3.0361 (0.0228) -0.1808 (0.0329) 18456.78 (69.96) 11268.42 (69.74) 1.592 (0.0032) 3.577 (0.2943) 454.1 (2.08)
integrated model 2.9366 (0.0163) -0.0989 (0.0074) 18113.81 (121.05) 11603.85 (122.77) 1.516 (0.0054) 0.879 (0.0201) 496.92 (0.77)
true value 5 -0.3 50 30000 1 1 500
naive model 3.0804 (0.027) -0.3768 (0.0336) 18175.54 (74.23) 11545.95 (73.97) 1.5939 (0.0032) 4.1281 (0.3787) 453.35 (1.29)
integrated model 2.9395 (0.0132) -0.2858 (0.004) 18058.62 (94.19) 11655.43 (95.21) 1.5166 (0.0046) 0.8667 (0.0213) 498.87 (0.69)
true value 5 -0.5 50 30000 1 1 500
naive model 3.1502 (0.0254) -0.7234 (0.0412) 17717.61 (82.68) 11998.41 (82.46) 1.5974 (0.0034) 3.9477 (0.3312) 454.13 (0.75)
integrated model 2.9937 (0.0129) -0.4811 (0.0042) 17680.24 (87.96) 12031.57 (88.86) 1.5063 (0.0047) 0.8291 (0.0251) 500.36 (0.78)
true value 5 -1 50 30000 1 1 500
naive model 3.5669 (0.043) -1.7535 (0.09) 15740.25 (113.86) 13964.36 (113.86) 1.5926 (0.0043) 6.4576 (0.5025) 471.3 (3.33)
integrated model 3.2607 (0.0177) -0.9702 (0.0074) 15650.42 (130.59) 14066.12 (132.08) 1.4461 (0.0057) 0.8718 (0.0259) 505.15 (0.98)
true value 5 -1.3 50 30000 1 1 500
naive model 4.0045 (0.1514) -2.2565 (0.5547) 13772.02 (120.38) 15929.25 (120.41) 1.5632 (0.0045) 12.2208 (2.356) 480.58 (3.11)
integrated model 3.1969 (0.3036) -0.7316 (0.5495) 13720.47 (126.43) 16004.76 (128) 1.3919 (0.0097) 2.9218 (2.2545) 514.38 (4.56)
true value 5 -1.5 50 30000 1 1 500
naive model 4.4603 (0.0723) -3.707 (0.1756) 12633.68 (117.4) 17067.83 (117.75) 1.5489 (0.0043) 12.0708 (0.7155) 485.18 (1.18)
integrated model 3.6474 (0.0193) -1.4779 (0.0129) 12443.71 (145.16) 17289.63 (147.48) 1.3512 (0.0055) 0.6374 (0.0466) 510.48 (1.31)

Table 2: Table for the detailed point estimates of all unknown parameters and their standard error;Sample size was from 20 to 300and the protein intensity located in the upper part of sigmoidal curve. Truncation was applied to the boundary of intensity level; the integrated model consistently yielded parameter estimates of α2 with similar or much less standard errors than the naïve model

Composite Element REF miRNA Corresponding genes MirTarbase MirTarbase  (Supported by strong experimental evidences) miRanda
Found by the integrated model only
p53-r-v hsa-mir-605 TP53     1 1  
n-cadherin-r-v hsa-mir-511-2 CDH2         1
akt-r-v hsa-mir-511-1 AKT1 AKT2 AKT3     1
eef2k-r-v hsa-mir-486 EEF2K         1
p53-r-v hsa-mir-181a-1 TP53         1
smad4-m-c hsa-mir-142 SMAD4         1
c-met-m-c hsa-mir-223 MET         1
caspase-8-m-c hsa-mir-541 CASP8         1
pcna-m-v hsa-mir-223 PCNA         1
n-cadherin-r-v hsa-mir-146a CDH2         1
syk-m-v hsa-mir-369 SYK         1
alpha-catenin-m-v hsa-mir-22 CTNNA1         1
smad3-r-v hsa-mir-142 SMAD3         1
chk1-r-v hsa-mir-605 CHEK1         1
er-alpha-r-v hsa-mir-181a-2 ESR1         1
aib1-m-v hsa-mir-605 NCOA3         1
rad50-m-c hsa-mir-22 RAD50         1
caspase-8-m-c hsa-mir-483 CASP8         1
syk-m-v hsa-mir-193a SYK         1
jnk2-r-c hsa-mir-511-1 MAPK9         1
p53-r-v hsa-mir-588 TP53         1
dj-1-r-c hsa-mir-145 PARK7         1
akt-r-v hsa-mir-142 AKT1 AKT2 AKT3     1
p27-r-v hsa-mir-205 CDKN1B         1
beta-catenin-r-v hsa-mir-485 CTNNB1         1
c-kit-r-v hsa-mir-21 KIT         1
chk1-r-v hsa-let-7b CHEK1         1
b-raf-m-na hsa-mir-511-1 BRAF         1
stat5-alpha-r-v hsa-mir-1224 STAT5A         1
igf-1r-beta-r-c hsa-let-7b IGF1R         1
pea-15-r-v hsa-mir-541 PEA15         1
beta-catenin-r-v hsa-mir-1228 CTNNB1         1
c-raf-r-v hsa-mir-1295 RAF1          
msh2-m-c hsa-mir-1247 MSH2          
msh2-m-c hsa-mir-199a-1 MSH2          
aib1-m-v hsa-mir-150 NCOA3          
bcl-2-m-v hsa-mir-1307 BCL2          
alpha-catenin-m-v hsa-mir-511-2 CTNNA1          
alpha-catenin-m-v hsa-mir-511-1 CTNNA1          
aib1-m-v hsa-mir-214 NCOA3          
bcl-2-m-v hsa-mir-652 BCL2          
ar-r-v hsa-mir-224 AR          
k-ras-m-c hsa-mir-150 KRAS          
ciap-r-v hsa-mir-223 BIRC2          
ciap-r-v hsa-mir-140 BIRC2          
ciap-r-v hsa-mir-511-2 BIRC2          
aib1-m-v hsa-mir-1228 NCOA3          
claudin-7-r-v hsa-mir-140 CLDN7          
ar-r-v hsa-mir-1228 AR          
bcl-xl-r-c hsa-mir-1295 BCL2L1          
ar-r-v hsa-mir-511-1 AR          
eef2k-r-v hsa-mir-181a-2 EEF2K          
tau-m-c hsa-mir-142 MAPT          
cdk1-r-v hsa-mir-1247 CDC2          
cdk1-r-v hsa-let-7a-3 CDC2          
cdk1-r-v hsa-let-7a-1 CDC2          
cdk1-r-v hsa-let-7a-2 CDC2          
eef2k-r-v hsa-mir-1908 EEF2K          
p27-r-v hsa-mir-136 CDKN1B          
eef2k-r-v hsa-mir-134 EEF2K          
setd2-r-na hsa-mir-511-2 SETD2          
smad1-r-v hsa-mir-483 SMAD1          
irs1-r-v hsa-mir-511-1 IRS1          
p21-r-c hsa-mir-150 CDKN1A          
ar-r-v hsa-mir-1307 AR          
bcl-2-m-v hsa-mir-1295 BCL2          
n-cadherin-r-v hsa-mir-142 CDH2          
smad4-m-c hsa-mir-1295 SMAD4          
cdk1-r-v hsa-let-7b CDC2          
smad1-r-v hsa-mir-139 SMAD1          
xiap-r-c hsa-mir-140 KDR          
claudin-7-r-v hsa-mir-511-2 CLDN7          
alpha-catenin-m-v hsa-mir-1228 CTNNA1          
bim-r-v hsa-mir-218-1 BCL2L11          
eef2k-r-v hsa-mir-758 EEF2K          
ar-r-v hsa-mir-452 AR          
ptch-r-c hsa-mir-25 PTCH1          
p27-r-v hsa-mir-758 CDKN1B          
c-kit-r-v hsa-mir-22 KIT          
smad1-r-v hsa-mir-377 SMAD1          
eef2k-r-v hsa-mir-217 EEF2K          
her2-m-v hsa-mir-511-1 ERBB2          
claudin-7-r-v hsa-mir-24-1 CLDN7          
aib1-m-v hsa-mir-139 NCOA3          
chk2-m-c hsa-mir-485 CHEK2          
claudin-7-r-v hsa-mir-758 CLDN7          
ar-r-v hsa-mir-589 AR          
chk2-m-c hsa-mir-140 CHEK2          
pcna-m-v hsa-mir-132 PCNA          
setd2-r-na hsa-mir-511-1 SETD2          
chk2-m-c hsa-mir-223 CHEK2          
alpha-catenin-m-v hsa-mir-605 CTNNA1          
claudin-7-r-v hsa-mir-431 CLDN7          
irs1-r-v hsa-mir-511-2 IRS1          
eef2k-r-v hsa-mir-519a-2 EEF2K          
yb-1-r-v hsa-mir-150 YBX1          
eef2k-r-v hsa-mir-136 EEF2K          
pcna-m-v hsa-mir-605 PCNA          
pcna-m-v hsa-mir-485 PCNA          
xrcc1-r-c hsa-mir-214 XRCC1          
pcna-m-v hsa-mir-199a-1 PCNA          
eef2k-r-v hsa-mir-432 EEF2K          
smad1-r-v hsa-mir-214 SMAD1          
eef2k-r-v hsa-mir-483 EEF2K          
c-kit-r-v hsa-mir-511-1 KIT          
foxo3a-r-c hsa-mir-142 FOXO3          
bim-r-v hsa-mir-485 BCL2L11          
snail-m-c hsa-mir-155 SNAI2          
er-alpha-r-v hsa-mir-877 ESR1          
stat5-alpha-r-v hsa-mir-605 STAT5A          
claudin-7-r-v hsa-mir-654 CLDN7          
claudin-7-r-v hsa-mir-337 CLDN7          
xrcc1-r-c hsa-mir-1247 XRCC1          
claudin-7-r-v hsa-mir-130b CLDN7          
mre11-r-c hsa-mir-766 MRE11A          
cdk1-r-v hsa-mir-140 CDC2          
bim-r-v hsa-mir-1976 BCL2L11          
gab2-r-v hsa-mir-654 GAB2          
her2-m-v hsa-mir-511-2 ERBB2          
chk2-m-c hsa-mir-24-1 CHEK2          
eef2-r-v hsa-mir-140 EEF2          
ar-r-v hsa-mir-1976 AR          
pr-r-v hsa-mir-31 PGR          
pkc-alpha-m-v hsa-mir-511-2 PRKCA          
mre11-r-c hsa-mir-511-2 MRE11A          
ku80-r-c hsa-mir-217 XRCC5          
ku80-r-c hsa-mir-24-1 XRCC5          
smac-m-v hsa-mir-1249 DIABLO          
claudin-7-r-v hsa-mir-370 CLDN7          
pcna-m-v hsa-mir-511-1 PCNA          
bim-r-v hsa-mir-409 BCL2L11          
eef2k-r-v hsa-mir-455 EEF2K          
53bp1-r-c hsa-mir-1295 TP53BP1          
xbp1-g-c hsa-mir-223 XBP1          
er-alpha-r-v hsa-mir-1276 ESR1          
dj-1-r-c hsa-mir-140 PARK7          
syk-m-v hsa-let-7e SYK          
er-alpha-r-v hsa-mir-1914 ESR1          
er-alpha-r-v hsa-mir-605 ESR1          
smad1-r-v hsa-mir-1247 SMAD1          
claudin-7-r-v hsa-mir-181b-2 CLDN7          
pr-r-v hsa-mir-629 PGR          
rab25-r-c hsa-mir-223 RAB25          
yb-1-r-v hsa-mir-1249 YBX1          
chk2-m-c hsa-mir-1295 CHEK2          
xrcc1-r-c hsa-mir-212 XRCC1          
gab2-r-v hsa-mir-1247 GAB2          
e-cadherin-r-v hsa-mir-1271 CDH1          
src-m-v hsa-mir-142 SRC          
p53-r-v hsa-mir-1908 TP53          
pkc-alpha-m-v hsa-mir-511-1 PRKCA          
notch3-r-c hsa-mir-365-2 NOTCH3          
p27-r-v hsa-mir-487a CDKN1B          
rab25-r-c hsa-mir-511-2 RAB25          
rab25-r-c hsa-mir-212 RAB25          
c-myc-r-c hsa-mir-511-2 MYC          
xiap-r-c hsa-mir-217 KDR          
smad1-r-v hsa-mir-212 SMAD1          
msh6-r-c hsa-mir-1249 MSH6          
ku80-r-c hsa-mir-1249 XRCC5          
smad1-r-v hsa-mir-487a SMAD1          
eef2k-r-v hsa-mir-370 EEF2K          
syk-m-v hsa-mir-411 SYK          
rad50-m-c hsa-mir-1295 RAD50          
eef2k-r-v hsa-mir-181b-2 EEF2K          
caspase-8-m-c hsa-mir-1247 CASP8          
gab2-r-v hsa-mir-337 GAB2          
bim-r-v hsa-mir-511-2 BCL2L11          
eef2k-r-v hsa-mir-522 EEF2K          
bim-r-v hsa-mir-145 BCL2L11          
p27-r-v hsa-mir-181b-1 CDKN1B          
msh6-r-c hsa-let-7a-2 MSH6          
msh6-r-c hsa-let-7a-3 MSH6          
ptch-r-c hsa-mir-223 PTCH1          
notch3-r-c hsa-mir-511-2 NOTCH3          
src-m-v hsa-mir-511-1 SRC          
jnk2-r-c hsa-mir-217 MAPK9          
bak-r-c hsa-mir-511-1 BAK1          
eif4e-r-v hsa-mir-605 EIF4E          
claudin-7-r-v hsa-mir-487b CLDN7          
p53-r-v hsa-mir-139 TP53          
p53-r-v hsa-mir-140 TP53          
igf-1r-beta-r-c hsa-mir-1228 IGF1R          
gab2-r-v hsa-mir-134 GAB2          
eef2k-r-v hsa-mir-369 EEF2K          
bcl-2-m-v hsa-mir-210 BCL2          
caveolin-1-r-v hsa-mir-551a CAV1          
paxillin-r-v hsa-mir-511-1 PXN          
yap-r-v hsa-mir-1295 YAP1          
yap-r-v hsa-mir-1228 YAP1          
smad1-r-v hsa-mir-485 SMAD1          
p53-r-v hsa-mir-589 TP53          
xiap-r-c hsa-mir-485 KDR          
e-cadherin-r-v hsa-mir-1306 CDH1          
pr-r-v hsa-mir-1908 PGR          
inpp4b-g-c hsa-mir-223 INPP4B          
smac-m-v hsa-let-7b DIABLO          
syk-m-v hsa-mir-539 SYK          
pcna-m-v hsa-mir-654 PCNA          
smad1-r-v hsa-mir-605 SMAD1          
syk-m-v hsa-mir-605 SYK          
smac-m-v hsa-mir-140 DIABLO          
eef2k-r-v hsa-mir-133a-1 EEF2K          
rab25-r-c hsa-mir-142 RAB25          
claudin-7-r-v hsa-mir-127 CLDN7          
akt-r-v hsa-mir-217 AKT1 AKT2 AKT3      
bak-r-c hsa-mir-511-2 BAK1          
aib1-m-v hsa-mir-1247 NCOA3          
cdk1-r-v hsa-mir-511-1 CDC2          
pea-15-r-v hsa-mir-433 PEA15          
smac-m-v hsa-mir-511-2 DIABLO          
gata3-m-v hsa-mir-99b GATA3          
inpp4b-g-c hsa-mir-150 INPP4B          
53bp1-r-c hsa-mir-589 TP53BP1          
c-kit-r-v hsa-mir-551a KIT          
mre11-r-c hsa-mir-511-1 MRE11A          
igfbp2-r-v hsa-mir-29c IGFBP2          
er-alpha-r-v hsa-mir-494 ESR1          
mek1-r-v hsa-mir-1228 MAP2K1          
erk2-r-na hsa-mir-511-1 MAPK1          
igf-1r-beta-r-c hsa-mir-658 IGF1R          
53bp1-r-c hsa-mir-146a TP53BP1          
alpha-catenin-m-v hsa-mir-1908 CTNNA1          
rad50-m-c hsa-mir-511-1 RAD50          
yap-r-v hsa-mir-217 YAP1          
bcl-2-m-v hsa-mir-203 BCL2          
p53-r-v hsa-mir-132 TP53          
caveolin-1-r-v hsa-mir-130b CAV1          
ku80-r-c hsa-mir-605 XRCC5          
syk-m-v hsa-mir-379 SYK          
c-kit-r-v hsa-mir-15b KIT          
gab2-r-v hsa-mir-181a-1 GAB2          
dj-1-r-c hsa-mir-342 PARK7          
bcl-x-r-c hsa-mir-1295 BCL2L1          
pcna-m-v hsa-mir-134 PCNA          
eef2k-r-v hsa-mir-130b EEF2K          
syk-m-v hsa-mir-22 SYK          
igf-1r-beta-r-c hsa-mir-203 IGF1R          
paxillin-r-v hsa-mir-766 PXN          
e-cadherin-r-v hsa-mir-329-1 CDH1          
pcna-m-v hsa-mir-493 PCNA          
smac-m-v hsa-mir-588 DIABLO          
smad1-r-v hsa-mir-455 SMAD1          
rad50-m-c hsa-mir-1247 RAD50          
inpp4b-g-c hsa-mir-766 INPP4B          
p70s6k-r-v hsa-mir-1247 RPS6KB1          
cdk1-r-v hsa-mir-605 CDC2          
smac-m-v hsa-mir-605 DIABLO          
smad1-r-v hsa-mir-1228 SMAD1          
alpha-catenin-m-v hsa-mir-214 CTNNA1          
Found by both the integrated model and the naïve model
notch3-r-c hsa-mir-150 NOTCH3     1 1 1
er-alpha-r-v hsa-mir-18a ESR1     1 1 1
p53-r-v hsa-mir-150 TP53     1 1  
beta-catenin-r-v hsa-mir-214 CTNNB1     1 1  
bim-r-v hsa-mir-181a-1 BCL2L11     1 1  
igf-1r-beta-r-c hsa-mir-223 IGF1R     1 1  
igf-1r-beta-r-c hsa-mir-139 IGF1R     1 1  
p27-r-v hsa-mir-181a-1 CDKN1B     1 1  
igf-1r-beta-r-c hsa-mir-145 IGF1R     1 1  
smad3-r-v hsa-mir-155 SMAD3     1 1  
msh6-r-c hsa-mir-21 MSH6     1 1  
caveolin-1-r-v hsa-mir-7-1 CAV1     1   1
bim-r-v hsa-let-7a-2 BCL2L11     1   1
bim-r-v hsa-let-7a-1 BCL2L11     1   1
bim-r-v hsa-let-7a-3 BCL2L11     1   1
n-cadherin-r-v hsa-mir-150 CDH2         1
yap-r-v hsa-mir-150 YAP1         1
er-alpha-r-v hsa-mir-766 ESR1         1
ku80-r-c hsa-mir-223 XRCC5         1
beta-catenin-r-v hsa-mir-223 CTNNB1         1
er-alpha-r-v hsa-mir-493 ESR1         1
claudin-7-r-v hsa-mir-214 CLDN7         1
e-cadherin-r-v hsa-mir-605 CDH1         1
claudin-7-r-v hsa-mir-1228 CLDN7         1
beta-catenin-r-v hsa-mir-511-1 CTNNB1         1
er-alpha-r-v hsa-mir-337 ESR1         1
er-alpha-r-v hsa-mir-299 ESR1         1
bim-r-v hsa-let-7b BCL2L11         1
syk-m-v hsa-mir-409 SYK         1
igf-1r-beta-r-c hsa-mir-142 IGF1R         1
e-cadherin-r-v hsa-mir-130b CDH1         1
pcna-m-v hsa-let-7a-2 PCNA         1
beta-catenin-r-v hsa-mir-511-2 CTNNB1         1
syk-m-v hsa-mir-654 SYK         1
e-cadherin-r-v hsa-mir-493 CDH1         1
eef2k-r-v hsa-mir-605 EEF2K         1
syk-m-v hsa-mir-337 SYK         1
snail-m-c hsa-mir-150 SNAI2         1
igf-1r-beta-r-c hsa-mir-511-1 IGF1R         1
e-cadherin-r-v hsa-mir-22 CDH1         1
er-alpha-r-v hsa-mir-323 ESR1         1
er-alpha-r-v hsa-mir-411 ESR1         1
bcl-2-m-v hsa-mir-22 BCL2         1
er-alpha-r-v hsa-mir-369 ESR1         1
e-cadherin-r-v hsa-mir-654 CDH1         1
beta-catenin-r-v hsa-mir-22 CTNNB1         1
n-cadherin-r-v hsa-mir-511-1 CDH2         1
caveolin-1-r-v hsa-mir-200a CAV1         1
chk2-m-c hsa-let-7b CHEK2         1
akt-r-v hsa-mir-223 AKT1 AKT2 AKT3     1
alpha-catenin-m-v hsa-mir-142 CTNNA1         1
igf-1r-beta-r-c hsa-mir-511-2 IGF1R         1
bim-r-v hsa-mir-1295 BCL2L11         1
e-cadherin-r-v hsa-mir-1908 CDH1         1
syk-m-v hsa-mir-150 SYK         1
e-cadherin-r-v hsa-mir-1228 CDH1         1
p70s6k-r-v hsa-mir-511-2 RPS6KB1         1
c-myc-r-c hsa-mir-486 MYC         1
b-raf-m-na hsa-mir-145 BRAF         1
eef2k-r-v hsa-mir-1247 EEF2K         1
igfbp2-r-v hsa-mir-664 IGFBP2         1
p70s6k-r-v hsa-mir-511-1 RPS6KB1         1
er-alpha-r-v hsa-mir-1910 ESR1         1
eef2k-r-v hsa-mir-487a EEF2K         1
claudin-7-r-v hsa-mir-493 CLDN7         1
e-cadherin-r-v hsa-mir-511-2 CDH1         1
yap-r-v hsa-mir-142 YAP1         1
akt-r-v hsa-mir-511-2 AKT1 AKT2 AKT3     1
alpha-catenin-m-v hsa-mir-223 CTNNA1         1
e-cadherin-r-v hsa-mir-218-1 CDH1         1
akt-r-v hsa-mir-766 AKT1 AKT2 AKT3     1
bim-r-v hsa-mir-605 BCL2L11         1
yap-r-v hsa-mir-605 YAP1         1
erk2-r-na hsa-mir-223 MAPK1         1
beta-catenin-r-v hsa-mir-146a CTNNB1         1
caspase-8-m-c hsa-mir-511-2 CASP8         1
syk-m-v hsa-mir-541 SYK         1
er-alpha-r-v hsa-mir-379 ESR1         1
chk2-m-c hsa-let-7a-3 CHEK2         1
chk2-m-c hsa-let-7a-2 CHEK2         1
e-cadherin-r-v hsa-mir-511-1 CDH1         1
eef2k-r-v hsa-mir-541 EEF2K         1
msh6-r-c hsa-mir-142 MSH6         1
chk2-m-c hsa-let-7a-1 CHEK2         1
e-cadherin-r-v hsa-mir-299 CDH1         1
b-raf-m-na hsa-mir-605 BRAF         1
c-met-m-c hsa-mir-511-2 MET         1
fak-r-c hsa-mir-616 PTK2         1
pkc-alpha-m-v hsa-mir-150 PRKCA         1
bim-r-v hsa-mir-223 BCL2L11         1
syk-m-v hsa-mir-483 SYK         1
msh6-r-c hsa-mir-146a MSH6         1
b-raf-m-na hsa-mir-223 BRAF         1
igfbp2-r-v hsa-mir-29b-1 IGFBP2         1
b-raf-m-na hsa-mir-511-2 BRAF         1
er-alpha-r-v hsa-mir-218-1 ESR1         1
syk-m-v hsa-mir-766 SYK         1
b-raf-m-na hsa-mir-193a BRAF         1
53bp1-r-c hsa-let-7b TP53BP1         1
caveolin-1-r-v hsa-mir-616 CAV1         1
rb-m-v hsa-mir-452 RB1         1
er-alpha-r-v hsa-mir-199a-2 ESR1         1
caveolin-1-r-v hsa-mir-200b CAV1         1
c-met-m-c hsa-mir-511-1 MET         1
jnk2-r-c hsa-mir-223 MAPK9         1
chk1-r-v hsa-mir-511-2 CHEK1         1
syk-m-v hsa-mir-511-2 SYK         1
igf-1r-beta-r-c hsa-mir-150 IGF1R          
beta-catenin-r-v hsa-mir-150 CTNNB1          
e-cadherin-r-v hsa-mir-214 CDH1          
53bp1-r-c hsa-mir-150 TP53BP1          
alpha-catenin-m-v hsa-mir-150 CTNNA1          
pr-r-v hsa-mir-150 PGR          
ku80-r-c hsa-mir-150 XRCC5          
claudin-7-r-v hsa-mir-766 CLDN7          
xrcc1-r-c hsa-mir-150 XRCC1          
e-cadherin-r-v hsa-mir-1247 CDH1          
e-cadherin-r-v hsa-mir-145 CDH1          
smad4-m-c hsa-mir-150 SMAD4          
igfbp2-r-v hsa-mir-224 IGFBP2          
eef2k-r-v hsa-mir-766 EEF2K          
e-cadherin-r-v hsa-mir-150 CDH1          
notch3-r-c hsa-mir-146a NOTCH3          
er-alpha-r-v hsa-mir-541 ESR1          
smac-m-v hsa-mir-150 DIABLO          
syk-m-v hsa-mir-145 SYK          
er-alpha-r-v hsa-mir-409 ESR1          
irs1-r-v hsa-mir-150 IRS1          
er-alpha-r-v hsa-mir-485 ESR1          
pr-r-v hsa-mir-22 PGR          
e-cadherin-r-v hsa-mir-766 CDH1          
syk-m-v hsa-mir-377 SYK          
claudin-7-r-v hsa-mir-1247 CLDN7          
syk-m-v hsa-mir-214 SYK          
er-alpha-r-v hsa-mir-134 ESR1          
er-alpha-r-v hsa-mir-758 ESR1          
e-cadherin-r-v hsa-mir-485 CDH1          
b-raf-m-na hsa-mir-150 BRAF          
syk-m-v hsa-mir-485 SYK          
er-alpha-r-v hsa-mir-214 ESR1          
dj-1-r-c hsa-mir-150 PARK7          
rad51-m-c hsa-mir-150 RAD51          
er-alpha-r-v hsa-mir-432 ESR1          
er-alpha-r-v hsa-mir-377 ESR1          
er-alpha-r-v hsa-mir-431 ESR1          
bim-r-v hsa-mir-193a BCL2L11          
gab2-r-v hsa-mir-766 GAB2          
er-alpha-r-v hsa-mir-487a ESR1          
msh6-r-c hsa-mir-1247 MSH6          
syk-m-v hsa-mir-140 SYK          
pcna-m-v hsa-mir-145 PCNA          
er-alpha-r-v hsa-mir-433 ESR1          
notch3-r-c hsa-mir-142 NOTCH3          
syk-m-v hsa-mir-487a SYK          
beta-catenin-r-v hsa-mir-766 CTNNB1          
rad50-m-c hsa-mir-150 RAD50          
pcna-m-v hsa-mir-1295 PCNA          
er-alpha-r-v hsa-mir-382 ESR1          
e-cadherin-r-v hsa-mir-134 CDH1          
er-alpha-r-v hsa-mir-370 ESR1          
alpha-catenin-m-v hsa-mir-766 CTNNA1          
c-myc-r-c hsa-mir-150 MYC          
e-cadherin-r-v hsa-mir-139 CDH1          
rb-m-v hsa-mir-150 RB1          
bcl-xl-r-c hsa-mir-150 BCL2L1          
er-alpha-r-v hsa-mir-483 ESR1          
msh6-r-c hsa-mir-214 MSH6          
msh6-r-c hsa-mir-1295 MSH6          
syk-m-v hsa-mir-127 SYK          
er-alpha-r-v hsa-mir-539 ESR1          
claudin-7-r-v hsa-mir-605 CLDN7          
her2-m-v hsa-mir-766 ERBB2          
her2-m-v hsa-mir-150 ERBB2          
claudin-7-r-v hsa-mir-485 CLDN7          
claudin-7-r-v hsa-mir-217 CLDN7          
c-myc-r-c hsa-mir-766 MYC          
chk2-m-c hsa-mir-145 CHEK2          
er-alpha-r-v hsa-mir-127 ESR1          
er-alpha-r-v hsa-mir-654 ESR1          
e-cadherin-r-v hsa-mir-409 CDH1          
syk-m-v hsa-mir-758 SYK          
er-alpha-r-v hsa-mir-136 ESR1          
syk-m-v hsa-mir-132 SYK          
e-cadherin-r-v hsa-mir-199a-1 CDH1          
igf-1r-beta-r-c hsa-mir-766 IGF1R          
e-cadherin-r-v hsa-mir-487a CDH1          
bim-r-v hsa-mir-150 BCL2L11          
ku80-r-c hsa-mir-766 XRCC5          
er-alpha-r-v hsa-mir-410 ESR1          
syk-m-v hsa-mir-493 SYK          
e-cadherin-r-v hsa-mir-199a-2 CDH1          
chk2-m-c hsa-mir-150 CHEK2          
er-alpha-r-v hsa-mir-1908 ESR1          
claudin-7-r-v hsa-mir-541 CLDN7          
53bp1-r-c hsa-mir-223 TP53BP1          
claudin-7-r-v hsa-mir-204 CLDN7          
eif4e-r-v hsa-mir-150 EIF4E          
p-cadherin-r-c hsa-mir-150 CDH3          
smad1-r-v hsa-mir-766 SMAD1          
pr-r-v hsa-mir-605 PGR          
xrcc1-r-c hsa-mir-223 XRCC1          
smad1-r-v hsa-mir-145 SMAD1          
igfbp2-r-v hsa-mir-452 IGFBP2          
53bp1-r-c hsa-mir-766 TP53BP1          
igf-1r-beta-r-c hsa-mir-589 IGF1R          
tau-m-c hsa-mir-150 MAPT          
igfbp2-r-v hsa-mir-150 IGFBP2          
e-cadherin-r-v hsa-mir-204 CDH1          
cdk1-r-v hsa-mir-150 CDC2          
pr-r-v hsa-mir-223 PGR          
pr-r-v hsa-mir-155 PGR          
syk-m-v hsa-mir-433 SYK          
syk-m-v hsa-mir-299 SYK          
e-cadherin-r-v hsa-mir-377 CDH1          
e-cadherin-r-v hsa-mir-212 CDH1          
chk1-r-v hsa-mir-150 CHEK1          
b-raf-m-na hsa-mir-1247 BRAF          
claudin-7-r-v hsa-mir-487a CLDN7          
syk-m-v hsa-mir-134 SYK          
e-cadherin-r-v hsa-mir-133a-1 CDH1          
rad50-m-c hsa-mir-223 RAD50          
dvl3-r-v hsa-mir-150 DVL3          
akt-r-v hsa-mir-150 AKT1 AKT2 AKT3      
p27-r-v hsa-mir-1276 CDKN1B          
e-cadherin-r-v hsa-mir-199b CDH1          
er-alpha-r-v hsa-mir-130b ESR1          
n-cadherin-r-v hsa-mir-224 CDH2          
claudin-7-r-v hsa-mir-1908 CLDN7          
e-cadherin-r-v hsa-mir-1295 CDH1          
syk-m-v hsa-mir-1295 SYK          
er-alpha-r-v hsa-mir-496 ESR1          
igf-1r-beta-r-c hsa-mir-1249 IGF1R          
syk-m-v hsa-mir-432 SYK          
lck-r-v hsa-mir-1269 LCK          
msh6-r-c hsa-mir-605 MSH6          
msh6-r-c hsa-mir-223 MSH6          
53bp1-r-c hsa-mir-1247 TP53BP1          
notch3-r-c hsa-mir-223 NOTCH3          
pr-r-v hsa-mir-142 PGR          
bak-r-c hsa-mir-150 BAK1          
e-cadherin-r-v hsa-mir-152 CDH1          
syk-m-v hsa-mir-1247 SYK          
dj-1-r-c hsa-mir-766 PARK7          
er-alpha-r-v hsa-mir-487b ESR1          
igfbp2-r-v hsa-mir-146a IGFBP2          
53bp1-r-c hsa-mir-1249 TP53BP1          
igf-1r-beta-r-c hsa-mir-146a IGF1R          
chk2-m-c hsa-mir-1247 CHEK2          
msh6-r-c hsa-mir-193a MSH6          
igfbp2-r-v hsa-mir-29a IGFBP2          
53bp1-r-c hsa-mir-511-2 TP53BP1          
eef2k-r-v hsa-mir-485 EEF2K          
53bp1-r-c hsa-mir-142 TP53BP1          
e-cadherin-r-v hsa-mir-337 CDH1          
xrcc1-r-c hsa-mir-140 XRCC1          
smac-m-v hsa-mir-212 DIABLO          
beta-catenin-r-v hsa-mir-1249 CTNNB1          
er-alpha-r-v hsa-mir-543 ESR1          
xrcc1-r-c hsa-mir-22 XRCC1          
foxo3a-r-c hsa-mir-511-1 FOXO3          
syk-m-v hsa-mir-410 SYK          
msh6-r-c hsa-mir-140 MSH6          
xiap-r-c hsa-mir-150 KDR          
claudin-7-r-v hsa-mir-218-1 CLDN7          
bim-r-v hsa-mir-1249 BCL2L11          
e-cadherin-r-v hsa-mir-181b-2 CDH1          
gab2-r-v hsa-mir-486 GAB2          
her2-m-v hsa-mir-145 ERBB2          
lck-r-v hsa-mir-149 LCK          
her2-m-v hsa-mir-212 ERBB2          
lck-r-v hsa-mir-1910 LCK          
53bp1-r-c hsa-mir-605 TP53BP1          
beta-catenin-r-v hsa-mir-1295 CTNNB1          
igf-1r-beta-r-c hsa-mir-1247 IGF1R          
paxillin-r-v hsa-mir-223 PXN          
er-alpha-r-v hsa-mir-145 ESR1          
yap-r-v hsa-mir-22 YAP1          
53bp1-r-c hsa-mir-511-1 TP53BP1          
yap-r-v hsa-mir-766 YAP1          
pr-r-v hsa-mir-1228 PGR          
igf-1r-beta-r-c hsa-mir-605 IGF1R          
e-cadherin-r-v hsa-mir-758 CDH1          
ku80-r-c hsa-mir-486 XRCC5          
smac-m-v hsa-mir-1247 DIABLO          
b-raf-m-na hsa-mir-214 BRAF          
her2-m-v hsa-mir-223 ERBB2          
eef2k-r-v hsa-mir-145 EEF2K          
syk-m-v hsa-mir-487b SYK          
syk-m-v hsa-mir-382 SYK          
eef2k-r-v hsa-mir-433 EEF2K          
e-cadherin-r-v hsa-mir-432 CDH1          
e-cadherin-r-v hsa-mir-382 CDH1          
syk-m-v hsa-mir-431 SYK          
beta-catenin-r-v hsa-mir-142 CTNNB1          
xbp1-g-c hsa-mir-150 XBP1          
er-alpha-r-v hsa-mir-1247 ESR1          
e-cadherin-r-v hsa-mir-181a-1 CDH1          
rad50-m-c hsa-mir-766 RAD50          
dvl3-r-v hsa-mir-511-2 DVL3          
bim-r-v hsa-mir-766 BCL2L11          
dvl3-r-v hsa-mir-511-1 DVL3          
smad3-r-v hsa-mir-22 SMAD3          
mek1-r-v hsa-mir-149 MAP2K1          
beta-catenin-r-v hsa-mir-224 CTNNB1          
chk2-m-c hsa-mir-214 CHEK2          
smad4-m-c hsa-mir-766 SMAD4          
e-cadherin-r-v hsa-mir-455 CDH1          
notch3-r-c hsa-mir-155 NOTCH3          
cd49b-m-v hsa-mir-150 ITGA2          
er-alpha-r-v hsa-mir-1228 ESR1          
claudin-7-r-v hsa-mir-134 CLDN7          
b-raf-m-na hsa-mir-766 BRAF          
notch3-r-c hsa-mir-452 NOTCH3          
smac-m-v hsa-mir-1295 DIABLO          
bim-r-v hsa-mir-455 BCL2L11          
er-alpha-r-v hsa-mir-99b ESR1          
e-cadherin-r-v hsa-mir-132 CDH1          
c-myc-r-c hsa-mir-223 MYC          
msh6-r-c hsa-mir-511-1 MSH6          
claudin-7-r-v hsa-mir-483 CLDN7          
er-alpha-r-v hsa-mir-486 ESR1          
xrcc1-r-c hsa-mir-193a XRCC1          
xrcc1-r-c hsa-mir-145 XRCC1          
gab2-r-v hsa-mir-605 GAB2          
claudin-7-r-v hsa-mir-377 CLDN7          
eef2k-r-v hsa-mir-539 EEF2K          
claudin-7-r-v hsa-mir-1295 CLDN7          
er-alpha-r-v hsa-mir-1224 ESR1          
p-cadherin-r-c hsa-mir-24-1 CDH3          
rab25-r-c hsa-mir-766 RAB25          
mek1-r-v hsa-mir-766 MAP2K1          
bim-r-v hsa-mir-214 BCL2L11          
claudin-7-r-v hsa-mir-1306 CLDN7          
eef2-r-v hsa-mir-150 EEF2          
msh6-r-c hsa-mir-511-2 MSH6          
syk-m-v hsa-mir-212 SYK          
syk-m-v hsa-mir-133a-1 SYK          
smad4-m-c hsa-mir-511-1 SMAD4          
igf-1r-beta-r-c hsa-mir-339 IGF1R          
er-alpha-r-v hsa-mir-329-1 ESR1          
chk2-m-c hsa-mir-605 CHEK2          
e-cadherin-r-v hsa-mir-433 CDH1          
smac-m-v hsa-mir-766 DIABLO          
chk2-m-c hsa-mir-766 CHEK2          
xrcc1-r-c hsa-mir-452 XRCC1          
alpha-catenin-m-v hsa-mir-1249 CTNNA1          
pr-r-v hsa-mir-224 PGR          
notch3-r-c hsa-mir-224 NOTCH3          
e-cadherin-r-v hsa-let-7e CDH1          
syk-m-v hsa-mir-199a-1 SYK          
alpha-catenin-m-v hsa-mir-1247 CTNNA1          
claudin-7-r-v hsa-mir-99b CLDN7          
xiap-r-c hsa-mir-223 KDR          
bim-r-v hsa-mir-486 BCL2L11          
caveolin-1-r-v hsa-mir-425 CAV1          
yap-r-v hsa-mir-511-2 YAP1          
claudin-7-r-v hsa-mir-539 CLDN7          
foxo3a-r-c hsa-mir-22 FOXO3          
dj-1-r-c hsa-mir-511-1 PARK7          
xrcc1-r-c hsa-mir-142 XRCC1          
e-cadherin-r-v hsa-mir-539 CDH1          
igfbp2-r-v hsa-mir-1249 IGFBP2          
igfbp2-r-v hsa-mir-142 IGFBP2          
smad3-r-v hsa-mir-511-2 SMAD3          
ku80-r-c hsa-mir-511-2 XRCC5          
smad1-r-v hsa-mir-150 SMAD1          
gab2-r-v hsa-mir-181a-2 GAB2          
stat5-alpha-r-v hsa-mir-218-1 STAT5A          
syk-m-v hsa-mir-381 SYK          
eef2k-r-v hsa-mir-150 EEF2K          
xrcc1-r-c hsa-let-7b XRCC1          
foxo3a-r-c hsa-mir-486 FOXO3          
pkc-alpha-m-v hsa-mir-223 PRKCA          
xiap-r-c hsa-mir-1247 KDR          
4e-bp1-r-v hsa-mir-1295 EIF4EBP1          
yap-r-v hsa-mir-511-1 YAP1          
ku80-r-c hsa-mir-1247 XRCC5          
b-raf-m-na hsa-let-7b BRAF          
xrcc1-r-c hsa-mir-146a XRCC1          
notch3-r-c hsa-mir-616 NOTCH3          
caveolin-1-r-v hsa-mir-182 CAV1          
nf2-r-c hsa-mir-511-1 NF2          
ar-r-v hsa-mir-511-2 AR          
claudin-7-r-v hsa-mir-125a CLDN7          
ku80-r-c hsa-mir-511-1 XRCC5          
claudin-7-r-v hsa-mir-1249 CLDN7          
n-cadherin-r-v hsa-mir-588 CDH2          
mek1-r-v hsa-mir-1249 MAP2K1          
yap-r-v hsa-mir-1247 YAP1          
stat5-alpha-r-v hsa-mir-455 STAT5A          
nf2-r-c hsa-mir-511-2 NF2          
xrcc1-r-c hsa-mir-217 XRCC1          
e-cadherin-r-v hsa-mir-370 CDH1          
vasp-r-c hsa-mir-504 VASP          
e-cadherin-r-v hsa-mir-541 CDH1          
notch3-r-c hsa-mir-22 NOTCH3          
bcl-xl-r-c hsa-mir-140 BCL2L1          
her2-m-v hsa-mir-1247 ERBB2          
gab2-r-v hsa-mir-377 GAB2          
p53-r-v hsa-mir-766 TP53          
er-alpha-r-v hsa-mir-154 ESR1          
rad50-m-c hsa-mir-217 RAD50          
er-alpha-r-v hsa-mir-1295 ESR1          
msh6-r-c hsa-mir-139 MSH6          
alpha-catenin-m-v hsa-mir-145 CTNNA1          
e-cadherin-r-v hsa-mir-486 CDH1          
p27-r-v hsa-mir-1254 CDKN1B          
syk-m-v hsa-mir-370 SYK          
pr-r-v hsa-mir-181a-1 PGR          
e-cadherin-r-v hsa-mir-891a CDH1          
er-alpha-r-v hsa-mir-181b-2 ESR1          
dj-1-r-c hsa-mir-1295 PARK7          
igf-1r-beta-r-c hsa-let-7i IGF1R          
smac-m-v hsa-mir-214 DIABLO          
claudin-7-r-v hsa-mir-133a-1 CLDN7          
paxillin-r-v hsa-mir-1249 PXN          
n-cadherin-r-v hsa-mir-140 CDH2          
53bp1-r-c hsa-mir-145 TP53BP1          
p-cadherin-r-c hsa-mir-23b CDH3          
xrcc1-r-c hsa-mir-511-1 XRCC1          
igf-1r-beta-r-c hsa-mir-1976 IGF1R          
beta-catenin-r-v hsa-mir-212 CTNNB1          
p27-r-v hsa-mir-541 CDKN1B          
er-alpha-r-v hsa-mir-1306 ESR1          
bcl-xl-r-c hsa-mir-145 BCL2L1          
smac-m-v hsa-mir-145 DIABLO          
irs1-r-v hsa-mir-24-1 IRS1          
syk-m-v hsa-mir-199a-2 SYK          
pr-r-v hsa-mir-766 PGR          
pea-15-r-v hsa-mir-511-2 PEA15          
syk-m-v hsa-mir-496 SYK          
bak-r-c hsa-mir-766 BAK1          
beta-catenin-r-v hsa-mir-605 CTNNB1          
e-cadherin-r-v hsa-mir-574 CDH1          
er-alpha-r-v hsa-mir-329-2 ESR1          
foxo3a-r-c hsa-mir-212 FOXO3          
eif4e-r-v hsa-mir-1249 EIF4E          
p27-r-v hsa-mir-99b CDKN1B          
igfbp2-r-v hsa-let-7f-1 IGFBP2          
notch3-r-c hsa-mir-1266 NOTCH3          
pea-15-r-v hsa-mir-511-1 PEA15          
mek1-r-v hsa-mir-296 MAP2K1          
dj-1-r-c hsa-mir-409 PARK7          
caveolin-1-r-v hsa-mir-375 CAV1          
53bp1-r-c hsa-mir-1228 TP53BP1          
53bp1-r-c hsa-mir-24-1 TP53BP1          
ku80-r-c hsa-mir-1228 XRCC5          
er-alpha-r-v hsa-mir-212 ESR1          
beta-catenin-r-v hsa-mir-217 CTNNB1          
bcl-xl-r-c hsa-mir-212 BCL2L1          
her2-m-v hsa-mir-217 ERBB2          
53bp1-r-c hsa-mir-22 TP53BP1          
p27-r-v hsa-mir-1307 CDKN1B          
bcl-2-m-v hsa-mir-511-2 BCL2          
notch3-r-c hsa-let-7i NOTCH3          
igfbp2-r-v hsa-let-7b IGFBP2          
notch3-r-c hsa-mir-511-1 NOTCH3          
er-alpha-r-v hsa-mir-139 ESR1          
lck-r-v hsa-mir-99b LCK          
53bp1-r-c hsa-mir-224 TP53BP1          
gab2-r-v hsa-mir-539 GAB2          
syk-m-v hsa-mir-543 SYK          
pten-r-v hsa-mir-766 PTEN          
pea-15-r-v hsa-mir-1910 PEA15          
xrcc1-r-c hsa-mir-224 XRCC1          
cdk1-r-v hsa-mir-217 CDC2          
caveolin-1-r-v hsa-mir-429 CAV1          
nf2-r-c hsa-mir-150 NF2          
gab2-r-v hsa-mir-455 GAB2          
chk2-m-c hsa-mir-139 CHEK2          
yap-r-v hsa-mir-1249 YAP1          
4e-bp1-r-v hsa-mir-145 EIF4EBP1          
gab2-r-v hsa-mir-541 GAB2          
e-cadherin-r-v hsa-mir-1249 CDH1          
mek1-r-v hsa-mir-504 MAP2K1          
claudin-7-r-v hsa-mir-891a CLDN7          
jnk2-r-c hsa-mir-150 MAPK9          
cdk1-r-v hsa-mir-664 CDC2          
gab2-r-v hsa-mir-1910 GAB2          
caveolin-1-r-v hsa-mir-96 CAV1          
setd2-r-na hsa-mir-223 SETD2          
igfbp2-r-v hsa-mir-223 IGFBP2          
4e-bp1-r-v hsa-mir-605 EIF4EBP1          
msh6-r-c hsa-mir-217 MSH6          
bcl-x-r-c hsa-mir-181a-1 BCL2L1          
b-raf-m-na hsa-mir-1249 BRAF          
igf-1r-beta-r-c hsa-mir-24-1 IGF1R          
e-cadherin-r-v hsa-mir-217 CDH1          
igfbp2-r-v hsa-mir-101-2 IGFBP2          
e-cadherin-r-v hsa-mir-431 CDH1          
n-cadherin-r-v hsa-mir-193a CDH2          
cdk1-r-v hsa-mir-511-2 CDC2          
ku80-r-c hsa-mir-142 XRCC5          
caveolin-1-r-v hsa-mir-210 CAV1          
xrcc1-r-c hsa-mir-766 XRCC1          
igf-1r-beta-r-c hsa-mir-486 IGF1R          
dj-1-r-c hsa-mir-22 PARK7          
caveolin-1-r-v hsa-mir-141 CAV1          
cdk1-r-v hsa-mir-29a CDC2          
e-cadherin-r-v hsa-mir-342 CDH1          
syk-m-v hsa-mir-1228 SYK          
beta-catenin-r-v hsa-mir-133a-1 CTNNB1          
nf2-r-c hsa-mir-129-2 NF2          
msh6-r-c hsa-mir-377 MSH6          
syk-m-v hsa-mir-584 SYK          
her2-m-v hsa-mir-1249 ERBB2          
chk2-m-c hsa-mir-133a-1 CHEK2          
dvl3-r-v hsa-mir-766 DVL3          
p27-r-v hsa-mir-486 CDKN1B          
gab2-r-v hsa-mir-485 GAB2          
xrcc1-r-c hsa-mir-365-2 XRCC1          
smad3-r-v hsa-mir-223 SMAD3          
e-cadherin-r-v hsa-mir-487b CDH1          
notch3-r-c hsa-mir-365-1 NOTCH3          
smac-m-v hsa-mir-1228 DIABLO          
caspase-8-m-c hsa-mir-99b CASP8          
irs1-r-v hsa-mir-223 IRS1          
alpha-catenin-m-v hsa-mir-212 CTNNA1          
gab2-r-v hsa-mir-431 GAB2          
cdk1-r-v hsa-mir-125a CDC2          
jnk2-r-c hsa-mir-140 MAPK9          
c-myc-r-c hsa-mir-217 MYC          
mek1-r-v hsa-mir-24-1 MAP2K1          
vegfr2-r-c hsa-mir-551a KDR          
setd2-r-na hsa-mir-140 SETD2          
rab25-r-c hsa-mir-217 RAB25          
xrcc1-r-c hsa-mir-139 XRCC1          
53bp1-r-c hsa-mir-214 TP53BP1          
lck-r-v hsa-mir-605 LCK          
gab2-r-v hsa-mir-487a GAB2          
er-alpha-r-v hsa-mir-381 ESR1          
ku80-r-c hsa-mir-146a XRCC5          
53bp1-r-c hsa-mir-217 TP53BP1          
smac-m-v hsa-mir-224 DIABLO          
er-alpha-r-v hsa-mir-140 ESR1          
msh6-r-c hsa-mir-574 MSH6          
bcl-x-r-c hsa-mir-212 BCL2L1          
n-cadherin-r-v hsa-mir-522 CDH2          
rad51-m-c hsa-mir-217 RAD51          
4e-bp1-r-v hsa-mir-193a EIF4EBP1          
notch3-r-c hsa-mir-505 NOTCH3          
igf-1r-beta-r-c hsa-mir-212 IGF1R          
foxo3a-r-c hsa-mir-140 FOXO3          
eef2k-r-v hsa-mir-223 EEF2K          
her2-m-v hsa-mir-584 ERBB2          
setd2-r-na hsa-mir-217 SETD2          
p27-r-v hsa-mir-144 CDKN1B          
Found by naïve model only
akt-r-v hsa-mir-146b AKT1 AKT2 AKT3 1    
chk1-r-v hsa-mir-511-1 CHEK1         1
alpha-catenin-m-v hsa-mir-140 CTNNA1         1
vasp-r-c hsa-mir-1276 VASP         1
paxillin-r-v hsa-mir-505 PXN         1
pten-r-v hsa-mir-511-2 PTEN         1
bcl-x-r-c hsa-mir-1276 BCL2L1         1
bim-r-v hsa-mir-132 BCL2L11          
claudin-7-r-v hsa-mir-1258 CLDN7          
aib1-m-v hsa-mir-511-2 NCOA3          
p21-r-c hsa-mir-25 CDKN1A          
aib1-m-v hsa-mir-511-1 NCOA3          
rad50-m-c hsa-mir-132 RAD50          
c-myc-r-c hsa-mir-541 MYC          
tau-m-c hsa-mir-144 MAPT          
ar-r-v hsa-mir-190 AR          
c-raf-r-v hsa-mir-190 RAF1          
alpha-catenin-m-v hsa-mir-217 CTNNA1          
c-myc-r-c hsa-mir-140 MYC          
rad51-m-c hsa-mir-140 RAD51          
er-alpha-r-v hsa-mir-150 ESR1          
lck-r-v hsa-mir-129-2 LCK          
aib1-m-v hsa-mir-190 NCOA3          
her2-m-v hsa-mir-139 ERBB2          
vasp-r-c hsa-mir-184 VASP          

Table 3: A list of miRNA/protein pairs suggested by the naïve model and the integrated model and they were classed into three groups: “Found by the integrated model only”, “Found by both the integrated model and the naïve model” and “Found by the naïve model only”. Pairs found by the integrated model were sorted by ascending order of adjusted p-values from the integrated model and the naïve model, respectively. Pairs found by the naïve model only were sorted by ascending order of adjusted p-values from the naïve model andintegrated model, respectively.In addition, a number “1” will mark under the column for pairs found by MirTarbase, MirTarBase with strong experimental evidences or miRanda, and those pairs will be list on the top of each group after pairs got ordered.

Discussion

The traditional way to detect direct targets of miRNA using miRNA-mRNA experiment method is limited, due to the fact that miRNAs may regulate their targets post-transcriptionally. In addition, other computational methods, which were based on optimal sequence complementarity of miRNA and mRNA, suffer from large percentage of false positives and of limited practical use. Taking the advantage of recent technique advance in measuring of miRNA expression and protein concentration levels in a high-throughput scale, we proposed to search for potential miRNA targets through a nonlinear hierarchical model. Computationally, this integrated model measures the correlation between miRNA and its targeting protein without making estimation of protein expression levels first as in the naïve method. We used both simulation studies and an application to the real data to compare our proposed method and the naïve method. Our simulation results suggested that both integrated and naïve methods can well control their type-I errors, however, the integrated method consistently showed higher detection powers than the naïve method under different scenarios, particularly when the protein intensity values were located close to the saturation point or the background noise level. In the real data example, our proposed integrated method detected much more potential miRNA targets than the naïve method. Furthermore, the number of potential miRNA targets, which can be confirmed by computational methods or literatures, is larger in the integrated method than that in the naïve method.

A significant association between a miRNA/protein pair can be either direct or indirect. For example, a miRNA may directly target and degrade a transcription factor (TF), which in turn induces indirect cascading effects of down-regulating the TF’s target genes. The association analyses from the simple or our integrated model would reveal both direct and indirect associations. In contrast, the other computer-based algorithms, e.g. miRanda, can only predict direct miRNA targets based on sequence comparison. In the real data analyses (Table 1), the relatively smaller percentage of overlap between our findings and miRanda database suggests that our algorithm may detect more indirect targets. This is sound since our algorithm is more powerful, as demonstrated by our simulation studies, and hence is capable of detecting smaller indirect associations. With the crossreference to miRanda database, those direct miRNA targets of more biological relevance could be filtered out to serve as top candidates for further biological validations. It is worth noting that our algorithm can indeed detect more direct miRNA targets in absolute number. Also, in Table 1, the results were based on a FDR of 10% for the multiple test adjustment; however, we also checked a FDR at 5% level and found the conclusion remained the same. That is, the proposed integrated method found more miRNA targets that appear in other existing databases, demonstrating its advantage over the naïve method.

Unknown parameters in our proposed model were estimated within the maximum likelihood framework. Using the asymptotic properties of maximum likelihood estimates, test statistics were straightforward to construct. However, some improvement can be made to further improve the proposed model. For example, we assumed a linear relationship between miRNA and protein to directly compare with the naïve method and to illustrate our model using simple examples, but in reality, the relationship between miRNAs and proteins could follow a nonlinear relationship, such as a dose-response curve. In this case, (Zi) can be replaced by other parametric or nonparametric functions. With some simple modifications, our model can be easily extended to relax these assumptions. Additionally, in this article the random error terms for different dilution steps were set to be independent and identically distributed as proposed in other RPPA analysis papers [10]. However, it is possible that the errors may be highly correlated. In this case, more complicated dependence matrix among serial dilution steps can also be readily incorporated into our model framework.

References

Select your language of interest to view the total content in your interested language
Post your comment

Share This Article

Relevant Topics

Recommended Conferences

Article Usage

  • Total views: 11509
  • [From(publication date):
    January-2015 - Aug 21, 2017]
  • Breakdown by view type
  • HTML page views : 7733
  • PDF downloads :3776
 

Post your comment

captcha   Reload  Can't read the image? click here to refresh

Peer Reviewed Journals
 
Make the best use of Scientific Research and information from our 700 + peer reviewed, Open Access Journals
International Conferences 2017-18
 
Meet Inspiring Speakers and Experts at our 3000+ Global Annual Meetings

Contact Us

 
© 2008-2017 OMICS International - Open Access Publisher. Best viewed in Mozilla Firefox | Google Chrome | Above IE 7.0 version
adwords