alexa HIV-PDI: A Protein-Drug Interaction Resource for Structural Analyses of HIV Drug Resistance: 1. Concepts and Associated Database
ISSN: 2157-7420
Journal of Health & Medical Informatics

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
All submissions of the EM system will be redirected to Online Manuscript Submission System. Authors are requested to submit articles directly to Online Manuscript Submission System of respective journal.
  • Research Article   
  • J Health Med Informat 2011, Vol 2(1): 104
  • DOI: 10.4172/2157-7420.1000104

HIV-PDI: A Protein-Drug Interaction Resource for Structural Analyses of HIV Drug Resistance: 1. Concepts and Associated Database

GHEMTIO Leo1*, SMAÏL-TABBONE Malika1, DJIKENG Appolinaire2, DEVIGNES Marie-Dominique1, KEMINSE Lionel3,4, KELBERT Patricia1, FOKAM Joseph3, MAIGRET Bernard1,4* and OUWE-MISSI-OUKEM-BOYER Odile3*
1Nancy University, LORIA, Orpailleur Team, Campus scientifique, BP 239, 54506 Vandoeuvre-lès-Nancy Cedex, France
2The J. Craig Venter Institute (JCVI), 9712 Medical Center Drive, Rockville, MD 20850, USA
3Centre International de Référence Chantal Biya (CIRCB) pour la Recherche sur la Prévention et la Prise en charge du VIH/SIDA, BP 3077, Yaoundé, Cameroon
4Harmonic Pharma, Espace Transfert, 615 rue du Jardin Botanique 54600 Villers-lès-Nancy, France
*Corresponding Author(s): GHEMTIO Leo, Nancy University, LORIA, Orpailleur Team, Campus Scientifique, BP 239, 54506 Vandoeuvre-lès-Nancy Cedex, France, Tel: +(33) 3 54 95 85 92, Fax: +(33) 3 83 27 83 19, Email: [email protected]
MAIGRET Bernard, Nancy University, LORIA, Orpailleur Team, Campus Scientifique, BP 239, 54506 Vandoeuvre-lès-Nancy Cedex, France, Email: [email protected]
OUWE-MISSI-OUKEM-BOYER Odile, Centre International De Référence Chantal Biya (CIRCB) Pour La Recherche Sur La Prévention Et La Prise En Charge Du VIH/SIDA, BP 3077, Yaoundé, Cameroon, Email: [email protected]

Received Date: Apr 11, 2011 / Accepted Date: Jul 10, 2011 / Published Date: Jul 12, 2011

Abstract

Overcoming the problem of resistance to antiretroviral drugs (ARVs) in HIV-infected patients is a major issue in AIDS research today. Advances in genome sequencing have facilitated the identification of a growing number of individual genotypes. Hence, it is now possible to understand HIV drug resistance at the molecular level by considering the three-dimensional (3D) structural interactions between ARVs and the mutated viral proteins of patients. Therefore, identification of the critical interactions lost further to one or several HIV mutations, and consequently the modifications of other molecular factors, could be indicators to propose appropriate ARVs escaping the resistance. This paper introduces the HIV-PDI (Protein-Drug Interactions) resource designed to be a decision making tool to propose alternative ARVs against a particular mutated viral protein, and thus to provide a personalized antiretroviral treatment. The HIV-PDI was conceived to serve as an integrated resource for studying HIV drug resistance at the structural level of the protein-drug interaction, with a special emphasis on the active site of the HIV drug target. As a first step, we focus on the well documented protease and related drugs. The HIV-PDI includes clinical information on patients, resistance to given ARVs treatments, HIV proteins structures and mutations, HIV protein/ARV drugs and their 3D interactions. The HIV-PDI may be queried using multiple combinations of fields including protein, drug and treatment conditions and coupled to visualization/analysis tools of 3D Protein-Drug interactions. The HIV-PDI resource can be used in order to help understand the appearance of resistance and to promote further novel drug and treatment developments based on analyses of 3D pattern of protein-drug interactions. A web-based version of HIV-PDI is available at http://hiv-pdi.loria.fr.

Introduction

Despite the availability of several antiretroviral drug molecules (ARVs), the rapid emergence of resistance to ARVs by the HIV virus has become a major obstacle to effective control of HIV infection and the treatment of infected patients [1,2]. Several retrospective and prospective studies have shown that the presence of a resistant strain of HIV before starting ARV treatment can often affect the outcome [3,4]. Thus, resistance to ARVs is now considered a top priority for HIV/ AIDS research and actions [5,6].

Many documented examples of resistance are directly related to mutations affecting the genetic make-up of the virus, and especially to those which relate to proteins targeted by ARVs. It is now well established that mutation-induced resistance involves structural modifications of viral proteins which reduce the ability of drug molecules to bind near the active site of the viral target but which do not otherwise impede the viral function. Indeed, X-ray crystallography studies have clearly shown that certain mutations can change the geometry of the viral protein active site, thus reducing the binding affinities of ARVs [7-10]. Consequently, the structural basis of ARV resistance can be understood by careful inspection of the interactions between the mutated target (i.e. the HIV variant) and its ligand (i.e. the drug) at the molecular level [11-13]. Therefore, to carry out molecular analyses of HIV disease mechanisms, it is necessary to augment current HIV-related databases with detailed three-dimensional (3D) protein and ligand structural information. Integrating and exploiting this knowledge could help to optimise the use of existing ARVs in first and second line treatment regimes [14,15].

Several HIV-related databases have been described for the collection and storage of information on known mutations, drug resistance, clinical data, and additional data on HIV proteins including correlated information for AIDS treatment [16-23] (see Table, Supplementary information 1, which shows the examples of commonly used HIV-related databases). For example, the Stanford HIV drug resistance database (HIVdb) [21] is a very well documented public database designed to represent, store, and analyze the variations in protein sequences which are responsible for HIV drug resistance. The HIVdb contains several hundred HIV-1 mutations in several flat files along with resistance data, results on drugs susceptibility, and clinical patient data. It also provides several tools for analyzing mutated viral sequences in order to help understand drug resistance [21,24]. Moreover, it has been used recently to support some very interesting studies on the relationship between HIV protein mutations and disease treatments in a given population [7,25,26].

Despite the availability of these resources linked to existing HIV databases [27], there is a need for bioinformatics tools which can relate information concerning viral mutations with 3D structural features in the corresponding proteins. Those mutations may affect the pattern of 3D interactions which characterize how a particular drug might bind to its cognate HIV protein. Therefore, identifying changes in such patterns could provide new insights into the molecular basis of HIV drug resistance. Our basic hypotheses are that (i) for a given patient, observed resistance to a treatment may rely on a loss of affinity of the delivered ARVs targeting HIV proteins, (ii) this loss of affinity is a consequence of structural modifications at the drug binding site due to mutations, and (iii) recovering lost affinity can be achieved, for example, by restoring a similar drug interaction pattern by identifying new interaction points in another class of ARVs to compensate for lost ones. Therefore, by using molecular modeling and protein-ligand docking, we presume that it will be possible to identify the main structural and interaction features leading to a loss of affinity and hence propose a possible route to overcome drug resistance. Therefore, in addition to being able to store and manipulate clinical and biological data, we explicitly designed our HIV-PDI database to include protein and ligand structural information, physico-chemical descriptors, and 3D protein-ligand interaction patterns. Nonetheless, for the initial version of our database, we decided to focus on the well-characterized and documented HIV protease and its associated inhibitors [28- 31] with the expectation that the lessons learned would be directly applicable to other HIV protein targets.

In this paper, we describe the design and implementation of our HIV-Protein Drug Interaction (HIV-PDI) database which is the centerpiece of a bioinformatics resource designed to integrate epidemiological, clinical, pharmacological, biological, chemical, and structural data collected from HIV infected patients and which allows close interaction with several protein structure analysis and visualizing tools for decision support.

Methods

Database requirements analysis

After extensive discussions with people on the ground, it became clear that in order to integrate current biological knowledge of viral proteins with clinical data, it would be desirable to create a repository which would provide unified access to a very diverse range of information drawn from equally diverse sources of data. For example, the repository would need to be able to store details on HIV protein sequences and their 3D structures, the chemical structures of known drug molecules and their physico-chemical properties, information about 3D protein-drug interactions, patient information including data on CD4 counts, viremia, drug treatment regimes, and treatment outcomes, for example. Furthermore, this data would need to be extracted from various textual sources such as scientific articles and patents, existing HIV databases (Supplementary information 1), as well as other protein sequence and structural databases such as the Protein Data Bank (PDB) [32], Swiss-Prot [33], GenBank [34], and various other data sources at the European Molecular Biology Laboratory (EMBL) [35], for example.

Our HIV-PDI database aims to span experimental, clinical, theoretical and computational domains related to HIV resistance and to provide access to all of these diverse resources via a user friendly graphical user interface (GUI) in an integrated manner which avoids information redundancy and facilitates data integrity. Furthermore, all relevant scientific publications and descriptions of the methods used to produce the data are also stored in our database. Hence, in case of inconsistent or contradictory data, the user has all of the necessary information to make his own interpretation. As illustrated (Figure 1), the HIV-PDI database can therefore be considered as a decision support system based on extensive collections of heterogeneous HIV related data.

health-medical-informatics-decision

Figure 1: Overview of the HIV-PDI decision support resource.

The HIV-PDI database can be queried easily on multiple fields to access data about ligands (e.g. ligand conformations or chemical groups), proteins (e.g. different protein conformation of protein structures), 3D protein-ligand interactions, as well as patients’ clinical data. For example, queries have been constructed which can search for examples of ligands which have a special chemical group, data on treatments of one or several patients, information about sequence mutations found in one or several patients, or data on ARV drug resistance observed in one or several patients. More complicated queries can also be formulated which can, for example, select ligands and proteins that are in complex and which have a given type of chemical interaction, provide the details of patient treatments which provoke mutations relating to drug resistance, identify treatment changes which have overcome previous cases of resistance, and select and visualize in 3D protein-drug interactions associated with mutations.

We expect that the clinical data, mutation and resistance information stored in the HIV-PDI database will be a very useful resource for finding relationships between patient genotype and drug resistance profiles using data mining techniques. Furthermore, the protein structure information and molecular descriptors [36-39] of the drugs stored in the HIV-PDI database will be useful for developing statistical models of drug potency and resistance.

The HIV-PDI conceptual data model

The HIV-PDI database was designed using the entity-relational (ER) database model to take into account the complexity of the data to be entered and subsequently extracted [40-42]. The ER approach was used to facilitate the processes of data requirement specification and conceptual database modeling. The resulting ER data model was then translated into a set of normalized relational tables. This phase was performed automatically using the publicly available DB Designer [43] computer-aided software engineering (CASE) tools. Details of the various data types and their relationships necessary to achieve our goal were integrated in a single ER conceptual model, as illustrated (Figure 2). The main entities in this model are described in further detail below.

health-medical-informatics-database

Figure 2: The ER diagram of the HIV-PDI database (entity attributes are not shown for clarity).

The entity Protein represents a wild type (WT) or a mutant protein sequence found in a patient. Each Protein instance is identified by a unique key (specific to this database) and is associated with at least one instance of the entity Patient. This association has a temporal descriptor to track the appearance or disappearance of a specific mutation in a patient according to the time-course of the patient’s treatment. The entity Protein is also associated with one or more instances of the entity Mutation because the sequence of a Protein instance may contain one or several mutations, and conversely because a single mutation can be related to several Protein instances. The current version of the database only stores information on the HIV protease, although future versions will also store information on other HIV protein targets. The entity Ligand represents instances of the ligands (i.e. experimental compounds or clinical drug candidates). This entity contains information characterizing the chemical and pharmaceutical properties of each compound stored.

To represent structural information about proteins and drugs, the entity Protein-MolecularStructure is related to the entity LigandMolecularStructure in order to define the 3D structures of protein-ligand complexes as well as the molecular interactions occurring within the complex. Wherever possible, crystallographic data from the PDB is used, but if this is not available the protein structure is modeled by homology and protein-ligand interactions are modeled using the Glide docking program [44]. A given Protein may be associated with more than one ProteinMolecular Structure because it can have more than one structural conformation. Similarly, a given Ligand may be associated with multiple LigandMolecular Structure instances, each of which represents a single conformation of the corresponding Ligand instance. Other descriptors that characterize protein-ligand complexes are also stored. These descriptors include affinity data (e.g. EC-50, KI, and IC-50) and the root mean squared deviation (RMSD) measure of their shape complementarities calculated from spherical harmonic (SH) analyses of their surface areas and volumes.

The entity Resistance represents the fact that a ligand no longer interacts optimally with a mutated form of the viral protease. Each instance of Resistance is thus associated with an instance of the Mutation entity which is itself documented by validated information from other databases or provided by patient data collected from various institutions. The relevant mutations can be obtained through the association between the entity Protein and the entity Mutation. The entity Patient represents instances of individual patients, and is associated with other entities which represent temporal clinical information such as the Treatment (drugs and regimen), Viremia and CD4 entities.

Data tables and data sources

The HIV-PDI database is implemented as a set of relational tables, as shown (Figure 3). Each main entity or relationship in the conceptual model typically maps to a relational table in the physical model. For example, the Ligand table contains several one-dimensional (1D) and two-dimensional (2D) attributes describing molecular properties such as compound name, chemical formula, pKa, hydrogen bond donor and acceptor information, along with other structural features. Each entry in this table is labeled according to its International Union of Pure and Applied Chemistry (IUPAC) International Chemical Identifier (InChiKey) [45]. An InChiKey identifier is a textual identifier for chemical substances, and is designed to provide a standard human readable way to encode molecular information and to facilitate chemical property searching in databases and on the internet. Nonetheless, a given ligand may be commonly known using several alternative names, and these are stored in the Ligand Synonym table. The Ligand Molecular Structure table stores the 3D molecular descriptors of drug structures such as atomic coordinates, SH molecular surface shape coefficients [46-48], molecular polarizability, and stereochemical state, for example. The Chemical Group table stores details of the chemical subgroups contained within each ligand compound.

health-medical-informatics-database

Figure 3: The main physical tables and their relationships in the HIV-PDI.

The data on compounds defined as HIV protease inhibitors was collected from four main sources. These are compounds approved as commercial drugs by the US Food and Drug Administration (FDA), patents, compounds undergoing clinical trials, compounds from the Life Chemical [49] database which are reported as HIV-related compounds, and HIV-related compounds described in articles published in various international peer-reviewed journals. The data on each compound identified from these sources is stored in the HIVPDI database. Additional 3D conformations and molecular descriptors were calculated and stored, using Omega [50] to generate multiple conformations for each compound, and by using the Chemaxon and InChI tools [45,51] to calculate the descriptors for each conformer. In a similar manner, the Protein table stores 1D and 2D information about HIV-related proteins, such as their family and sub-family types, and genetic and amino acid sequences. The Protein Molecular Structure table stores 3D protein structure information, such as the atomic coordinates of all of the amino acids in a protein, the SH coefficients of a protein's surface shape and the shape of its ligand binding site. If a protein is known using several synonyms, these may be stored in the Protein Synonym table.

The main sources of protein information are Swiss-Prot [33] and the PDB. The WT sequence of HIV-1 was collected from the Stanford HIVdb. This sequence has a Swiss-Prot identifier of Q9WFL7 and the PDB code is 2QMP. As is usual practice, this WT sequence is taken as the reference sequence against which all mutations are defined. When no X-ray structure for a given mutation is available in the PDB, a 3D model of the corresponding sequence was built by homology using the 2QPM structure as a starting template in the Modeler program [52,53]. The experimental and model-built protein structures were then used as starting structures for molecular dynamics (MD) simulations which were used to generate multiple conformational samples of each protein. Each MD run consisted of a 1 ns simulation in explicit solvent, as calculated by the NAMD software [54]. This produced a set of conformational states for each protein for subsequent 3D shape analyses and protein-ligand docking simulations. The MSSH program was used to compute SH-based descriptors of the volumes and surfaces of each protein conformation [46,55].

The Protein Molecular Structure_has_Ligand Molecular Structure table stores inform-ation concerning protein-drug complexes, such as affinity data (EC-50, KI, and IC-50), lists of 3D interactions, and the RMS difference between the SH shapes of the protein and ligand surfaces [46]. Data on protein-drug interactions were extracted from the 3D structures of known HIV protease-drug complexes. The affinity of protein-drug complexes was extracted from the corresponding scientific literature. For compounds for which no protein-ligand structure is available in the PDB, flexible protein-ligand docking was performed using the Glide program [44,45,56-58] in order to generate 3D structures of complexes between each ligand and the WT and mutated HIV proteases. Lists of residues interacting with each ligand were extracted from the modeled complex using the LPC software [59]. The SH coefficient RMSD (quality of fit) between each protein pocket and ligand molecule was calculated using the SHEF program [55].

The Patient table stores information relating to individual patients such as their name, sex, and date of birth. The CD4 table stores data on the CD4 T-lymphocyte counts recorded for individual patients (cells/ μL, measurement date). The Viremia table stores data on the plasma viral load of patients (copies of viral genome/mL, measurement date). Similarly, the Treatment and Treatment Break tables store data on the patients’ treatments, such as the date of the treatment, the treatment specificity, details on interruption schedules and any clinical remarks. The Mutation table stores data on each patient’s HIV mutations, such as the WT position and type of the mutated amino acid and the corresponding mutated codons. The Resistance table stores details of the compounds for a patient found to have a resistant strain of the virus.

The clinical data on HIV-infected patients was extracted from the Stanford HIVdb. These data are related by a unique identification number for each patient. The list of known mutations was extracted from the available literature and from the Stanford HIVdb, and from the Los Alamos, Swiss-Prot, PDB, and International AIDS Society-USA (IAS-USA) [60] databases [61,62] (Supplementary information 1). The full sequences of mutated proteins were deduced by comparison with the reference sequence of the WT HIV protease.

Results

Database implementation

The relational database is implemented in PostgreSQL [63] (version 8.3.7) on a 64-bit Linux operating system running the Apache web server. Python wrapper scripts were used to import data from external data sources into the database. The Structured Query Language (SQL) [40,64] is used to retrieve and manage the data stored in the HIV-PDI database. For users who are not familiar with using SQL, a web-based graphical user interface (GUI) has been implemented using Python and the object-oriented web application framework CherryPy [65].

Database statistics

Table 1 lists some overall statistics on the number of records currently stored in the HIV-PDI database. Currently, the Ligand table contains a total of 5,962 non-redundant entries concerning HIVrelated compounds. This data can be extracted using a SQL query such as “SELECT PatentInChiKey, HBDonor, TautomerNumber, AtomNumber, LogP, PolarSurfaceArea, Stage, Class, DrugFunction, Diseasename, Name FROM Ligand;”. Table 2 shows a partial listing of the results of this query.

  Ligand Protein Patient Mutation Viremia Treatment
Number of entries 5,962 2,493 2,029 7,791 28,447 13,629

Table 1: A summary of the total number of entries in the HIV-PDI database, version 1.0.

Inchi key Hb donor Tautomer number Atom number Logp Polar surface area Stage Name Logd
YMARZQAQMVYCKC-RKPZHZFKBN 3 2 70 2.38 139.57 Commercial FDA-Agenerase 2.5
NZPXPXAGXYTROM-YKPJHVMUBZ 0 10 75 7.14 114 Commercial FDA-Aptivus 6.2
QDRMCFDXPIEYGX-XSBXPMGLBM 1 162 133 0.72 195.25 Commercial FDA-Atripla 3
LHCOVOKZWQYODM-AKUSELONBT 2 9 58 -2.47 108.3 Commercial FDA-Combivir -3.4
ZZHIRSCAVGFBKE-UHFFFAOYAA 0 3 23 1.02 45.76 R&D LC-PR-1 1.1
CBVCZFGXHXORBI-IXYYIQBABV 4 9 92 2.58 118.03 Commercial FDA-Crixivan 2.1
XQSPYNMVSIKCOC-KIJMNMGRBZ 2 3 26 -1.31 113.45 Commercial FDA-Emtriva 0.1
JTEGQNOMFQHVDC-ZCIVMJCZBB 2 3 26 -1.27 113.45 Commercial FDA-Epivir -0.3
MBFKCGGQTYQTLR-ZUCZGPQLBE 3 6 46 0.62 101.88 Commercial FDA-Epzicom 0.4
WREGKURFCTUGRC-KSLURUABBM 2 3 28 -1.23 88.15 Commercial FDA-Hivid -0.4
PYGWGZALEOIKDF-UHFFFAOYAM 2 6 43 5.24 120.64 Commercial FDA-Intelence 5.2
ZEJGORPKXHJSER-FYWRMAATBT 2 44 53 0.14 152.85 Commercial FDA-Isentress -2.9

Table 2: Sample selection from the Ligand table.

The Protein table contains 2,493 entries. This table may be listed with the SQL query “SELECT PatentidProtein, Patient_idPatient, ProteinSequence, GeneSequence, Family, Type2, Souche FROM Protein;” (see Table 3). The Patient table holds 2,029 entries. It may be queried using “SELECT idPatient, Alias, DateBirth, PlaceBirth, Sex, Address FROM Patient;”. The Mutation table contains 7,791 entries. It may be queried using: “SELECT PatentidMutation, AAPosition, Reference, WildTypeAA, MutantAA, DrugClass, Compound FROM Mutation;”. The Treatment table holds 13,629 records. It may be queried using “SELECT idTreatment, Patient_idPatient, DateTreatment, Treatment, Comment_2 FROM Treatment;” (see Table 4).

Id protein Id patient Protein sequence Strain Family Type
St-20245-0-I13V-L63A 20245 PQITLWQRPLVTVKIGGQLKEAL
LDTGADDTVLEE
MNLPGRWKPKMIGGIGGFIK
VRQYDQIAIEICGHKAIGTVLVGPT
PVNIIGRNLLTQIGCTLNF
Unknown Protease HIV-1
St-19227-0-P39S-L63P-A71T 19227 PQITLWQRPLVTIKIGGQ
LKEALLDTGADDTVLEEMNLSGRWKPK
MIGGIGGFIKVRQYDQIPIEICGHKTIGTV
LVGPTPV NIIGRNLLTQIGCTLNF
Unknown Protease HIV-1
Sw-O41798   PQITLWQRPLVTVKIGGQLIEALLDTGADDTVLEGI
NLPGKWKPKMIGGIGGFIKVRQYDQILIEIGGKKA
IGTVLVGPTPINIIGRNMLTQIGCTLNF
M Protease HIV-1
Sw-O89290   PQITLWQRPLVTIRVGGQLKEALLDTGADDTVLEDVN
LPGKWKPKMIGGIGGFIKVKQYDSILIEICGHRAIGTV
LVGPTPVNIIGRNMLTQIGCTLHF
M Protease HIV-1

Table 3: Sample selection extracted from the protein table.

Id treatment Id patient Date treatment Treatment Comment
29440_AZT_-359 29440 -359 AZT NOTHING
29440_DDI_-162 29440 -162 DDI NOTHING
29440_AZT_0 29440 0 AZT NOTHING
29440_DDI_0 29440 0 DDI NOTHING
29440_3TC_0 29440 0 3TC NOTHING
20245_AZT_0 20245 0 AZT NOTHING
20245_3TC_0 20245 0 3TC NOTHING
20246_D4T_0 20246 0 D4T NOTHING
20246_D4T_23 20246 23 D4T NOTHING
20246_3TC_23 20246 23 3TC NOTHING
20247_D4T_0 20247 0 D4T NOTHING
20247_3TC_0 20247 0 3TC NOTHING
20248_D4T_0 20248 0 D4T NOTHING
20249_AZT_0 20249 0 AZT NOTHING
20249_3TC_0 20249 0 3TC NOTHING
20250_AZT_0 20250 0 AZT NOTHING
20250_3TC_0 20250 0 3TC NOTHING
20251_AZT_0 20251 0 AZT NOTHING
20251_3TC_0 20251 0 3TC NOTHING
20252_D4T_0 20252 0 D4T NOTHING

Table 4: Sample selection extracted from the treatment table.

Graphical user interface

The main page of the graphical user interface (GUI) (see Figure 4, which shows the screenshot of the main GUI page for accessing all fields in the HIV-PDI) allows access to all of the data present in the database using several main request categories. All of the database entities may be explored using this page. For example, the result of a request to visualize the data stored for patients (see Figure 5, which shows the screenshot of the GUI for patient’s data in the HIV-PDI) presents additional links to view the data in more detail (1D, 2D, and 3D visualizations are available).

health-medical-informatics-screenshot

Figure 4: The screenshot of the main GUI page for accessing all fields in the HIV-PDI.

The database may also be queried directly by requests on entities such as Mutation (see Figure 5), Ligand, Protein or Patient, for example. More complex queries are also possible by using a combination of requests on different entities. Because all of the entity levels are linked to each other, the GUI pages allow the user to traverse easily from patient data to molecular structural information or from protein sequences and therapies to drug resistance and treatments, for example.

health-medical-informatics-data

Figure 5: The screenshot of the GUI for patient’s data in the HIV-PDI.

Discussion

An important aspect of the present work is that a conceptual data model was employed to link data and information from diverse domains throughout the database design process. By transferring clinical and biological user requirements to a graphical data model, the conceptual design approach allowed all relevant data and relationships to be considered at a glance. The resulting tables that make up the relational database contain data on clinical and experimental measurements, 3D structural data relating to proteases and anti-protease drugs, along with and physico-chemical properties and molecular descriptors of known drug molecules and drug candidates.

Taken together, this integrated resource now makes it possible to perform a wide range of database queries which will be useful for helping to understand the structural causes of ARV resistance phenomenon observed in patients. To our knowledge, the HIV-PDI is the first HIV-related database and resource which fully integrates clinical and biological data on HIV proteins with 3D structural and physicochemical information on ARV drugs and drug candidates. Overall, the HIV-PDI database supports patient-oriented queries, efficient follow up of patient treatments, convenient reviews of treatment outcomes, and systematic documentation of ARV drug resistance. It also brings the promise of being able to identify alternative treatments for observed cases of resistance. Hence one ultimate goal of our HIV-PDI resource is to support an integrated bioinformatics platform to provide a decision support tool to help clinicians making informed choices for second line treatments against HIV in patients presenting resistance. Furthermore, this platform could also be used for identifying and designing new ARVs which can address the on-going problem of increasing viral resistance to existing therapies.

Conclusions

We have designed and implemented the HIV-PDI database to relate the problems of ARV drug resistance in patients to protein-ligand interactions at the molecular level. The HIV-PDI database therefore provides a useful repository for the collection and interpretation of diverse information which could be crucially important in understanding HIV disease processes at the molecular level and in proposing alternative ARV therapies on a patient-specific basis. The introduction of 3D structural analysis about the protein/ ligand complexes in a database related to HIV and the consequences of mutations on the stability of the protein/ligand complexes make the main difference with the others known databases. A subsequent paper will present examples of using the HIV-PDI resource in the context of clinical decision support for dealing with patients presenting resistance to current ARV therapies.

Future developments of HIV-PDI resource will include adding data for additional HIV protein targets and drugs, linking out to more chemoinformatics modules to provide richer and more sensitive analyses of protein-ligand interactions and to perform more sensitive comparisons of ligand physico-chemical properties. We will also provide links to knowledge extraction modules to predict virological responses to therapy or resistant mutations [10,11,27,66-68]. Therefore, our database platform could open new avenues for identifying and evaluating new drugs which can bind to specific mutated HIV proteins with high affinity and thus provide new and more tailored treatments for ARV-resistant HIV mutations.

Authors’ contributions

All authors have jointly developed the research concept and collaborated on the writing of the manuscript. As the main author Leo GHEMTIO has initiated the study, carried out the computational analyses, has interpreted the results, and drafted the manuscript. All authors revised the manuscript and approved its final version.

Acknowledgements

The authors thank the Bill & Melinda Gates Foundation for financial support through the Grand Challenge Exploration grant N° 52034 (Round I). They are grateful to Michel Souchet, Dave Ritchie, Birama Ndiaye, and Florent Petronin for their respective contribution to the present work.

Leo Ghemtio was supported by grants from INRIA (Institut National de Recherche en Informatique et en Automatique), CNRS (Centre National pour la Recherche Scientifique) and the Bill & Melinda Gates Foundation;

Lionel Keminse and Joseph Fokam were supported by grant from the Bill & Melinda Gates Foundation.

We thank Openeye and Chemaxon for providing academic licences for their software.

This work was also supported in part by Region Lorraine within the framework of the PRST MISN (MBI operation).

References

Citation: GHEMTIO L, SMAÏL-TABBONE M, DJIKENG A, DEVIGNES MD, KEMINSE L, et al. (2011) HIV-PDI: A Protein-Drug Interaction Resource for Structural Analyses of HIV Drug Resistance: 1. Concepts and Associated Database. J Health Med Informat 2:104. Doi: 10.4172/2157-7420.1000104

Copyright: © 2011 GHEMTIO L, 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.

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

Post Your Comment Citation
Share This Article
Article Usage
  • Total views: 13031
  • [From(publication date): 7-2011 - Aug 24, 2019]
  • Breakdown by view type
  • HTML page views: 9211
  • PDF downloads: 3820
Share This Article
Leave Your Message 24x7
Top