Computational modeling of protein-ligand interactionsvorgogoz/BioInfoCourses/2018...Computational...
Transcript of Computational modeling of protein-ligand interactionsvorgogoz/BioInfoCourses/2018...Computational...
Computational modeling of
protein-ligand interactionsO. Taboureau, AC. Camproux, P. Tuffery
UMR-S 973
Club Bioinfo Ocotber 11, [email protected]
Molecules Therapeutiques in Silico (MTi) Unit
http://www.mti.univ-paris-diderot.fr/
Peptide
design
Profiling
Virtual
screening
& ADMET
Team 3
M.Miteva
B.Villoutreix
Team 2
AC. Camproux
O. Taboureau
Team 1
P. Tuffery
UMR-S 973
(Dir : B. Villoutreix)
2014-2018
Computational modeling of protein-ligand interactions
2019-2023
Computational modeling of protein-ligand interactions
Plate-forme RPBS (IBiSA) (P. Tufféry, S. De Vries, J. Rey)
(Ressource Parisienne en Bioinformatique Structurale)
PAAS-SdV
http://bioserv.rpbs.univ-paris-diderot.fr/index.html
Targets
Proteins
Protein-protein
Interactions
ADN, ARN, ...
Molecules
Small compounds
Peptides
AntibodiesMulti targets profiling
Pathway and ppi
Clinical effects
Gene expression
and toxicogenomic
Genetic variations
2 research areas :
- Pharmacological profiling
- Protein-peptide interactions
and peptides design
Genomic information
Computational modeling of proteine-ligand interactions
Objectives
Peptide design: Recognition
bioserv.rpbs.univ-paris-diderot.fr/services/BCSearch Guyon et al., Bioinformatics 2014; NAR 2015
Input: gapped amino acid sequence
Output: 3D fragments matching flanks & size 3D
Fast geometric search in protein structures Accurate loop modeling
Karami et al., Sci. Rep. 2018
Over 50% of successful loop models are derived from unrelated proteins, indicating that fragments under
similar constraints tend to adopt similar structure,
beyond mere homology.
Peptide design: Modeling
Forward-backtrack, K-best, Taboo sampling3D encoding of structures
Using Hidden Markov Models
Protein binding site identification Folding peptide at protein binding site
http://bioserv.rpbs.univ-paris-diderot.fr/services/PEP-SiteFinder/
http://bioserv.rpbs.univ-paris-diderot.fr/services/PEP-FOLD/
http://bioserv.rpbs.univ-paris-diderot.fr/services/PEP-FOLD3/
Maupetit et al., NAR, 2009
Thevenet et al., NAR, 2012
Shen et al., J. Chem. Theor. Comput., 2014
Lamiable et al. NAR, 2016
Lamiable et al, J. Comput Chem., 2016
Saladin et al, Nucleic Acids Res., 2014
Camproux et al. J. Mol. Biol., 2004
Peptide structure modeling
Protein-protein interactions
Peptide discovery: BactPepDB: a database of predicted peptides in prokaryotic genomes
Rey et al., Database, 2015
Peptide design, off-target: Non sequential alignments
èFar more difficult problem but useful for protein surface comparisons. atoms involved in a
function, : interaction with a drug, interaction with other proteins
ARA H201
10 20 30
40 50 60
MLTILVALAL FLLAAHASAR QQWELQGDRR
CQSQLERANL RPCEQHLMQK IQRDEDSYER
70 80 90
100 110 120
DPYSPSQDPY SPSPYDRRGA GSSQHQERCC
NELNEFENNQ RCMCEALQQI MENQSDRLQG
130 140 150
RQQEQQFKRE LRNLPQQCGL RAPQRCDLDV ESGGRDRY
l We use graph theory : search for cliques or quasi-cliques in product graphs
l We developped a similarity measure already used in image analysis for face or object recognition
(Binet-Cauchy Kernel)
Rasolohery I. et al., J. Chem. Inf. Model. 2017
Comparison of atom positions independently of the amino-acid sequence order
Membrane proteins motions
Molecular DynamicSimutations
Coarses grain
Synthetic channels
Pharmacological profiling: Target based
Conclusions & perspectives
Pockdrug extension using dynamic molecular simulations:
Characterization and validation of proteins as therapeutic targets and prediction of
drug-target interactions based on statistical & bioinformatics structural approaches
Prediction of ligand profiles using biostatistical approaches
Further deepen the profiling study: propose ligand candidates
that can bind with high affinity druggable pockets
Current applications: NS1 (non structural protein of influenza virus)
Proteins flexibility and transient pocket analysis
coA catalytic pocket
Prediction of druggable targets:
Proteins able to bind drug-like molecules, using
statistical approaches: Pockdrug websiteBorrel et al, 2015, Hussein et al, 2015
Hussein et al, 2017, Regad et al, 2017, Alam et al, 2018 Cerisier et al, 2017a, Cerisier et al, 2017b
Inhibition of HIV-2 protease: study of the resistance mechanism using in silico
approaches
LS
3D structures : 15 PR1
13 PR2unbound, bound forms
(indinavir, darunavir, amprenavir)
Tâche 1 : Récolte des
données sur PR1, PR2 et les IPs
Tâche 2 : Analyse et
comparaison de la structure des PR1 et
PR2
Tâche 3 : Analyse
et comparaison des IPs
Tâche 4 : Analyse et
comparaison des modes de liaison (=
site de liaison + IPs)
Construire des modèles statistiques
pour prédire l’activité d’une molécule vis à vis de PR2 sauvage
ou mutante
Mieux comprendre les mécanismes
de la résistance naturelle et acquise du VIH-1 et VIH-2
Tâche 5 : Prédiction
des effets des mutations de
résistance sur PR2
SA-Conf
AA
LS
Triki et al, 2018a, Triki et al, 2018b, Triki et al, 2018c, Triki et al, 2018d accepted
Pharmacological profiling: Target based – Application on HIV-2 proteases
Projet VIH2-Bioinfo
Pharmacological profiling: Data integration
Chemical
Bioactivity
Disease Pathway
Gene Ontology
Therapeutic
effect(ATC code)
Disease
complex network
Structural chemical
similarity and QSAR
Sequences protein similarity
Disease
Adverse drugreaction
Fingerprints, SEA
BLASTProtein
Integration of genetic variation (SNPs) -pharmacogenomics
Integration of gene expression
Protein-ADR prediction
Drug-target prediction
Enrichment analysis
http://potentia.cbs.dtu.dk/ChemProt/ Kringelum J. Database 2016
Promiscuity
ChemProt: focused on Drug-Target-Biological outcomes profiling
Many data not always linked together Need of data integration
Pharmacological profiling: Systems chemical toxicology
Toxicogenomics data analysis
More than 10000samples
WY-14643
Gene pvalNom(invitro)
pvalNom(invivo)
Acot1 4.45e-04 6.74e-08
Aig1 2.82e-04 2.14e-06
Ehhadh 4.43e-05 3.18e-06
Acox1 5.31e-04 4.97e-06
Cpt1b 1.53e-03 5.95e-06
Cpt2 2.09e-03 6.82e-06
Slc27a2 6.04e-04 1.04e-05
Fatty acid transporter
Lipid metabolism
Acetaminophen
Gene Set Enrichment Analysis approach (GSEA)
Com
pounds
Genes
Analysis of large scale microarray data
Taboureau O. J. Pharmacogenom. Pharmacoprot. 2012
Ligand-protein interaction Cell deregulation Metabolic pathways affected
+ + =- Suggestion of gene’s
biomarkers (ALDH7A1,
OSBL9 …)
- Suggestion of molecules
with a risk.
Such analysis can be used for computational phenotypic screen.
Prediction of DILI compounds and genes associated to steatosis
Pharmacological profiling: Example with Steatosis and DILI
Aguayo-Orozco et al., Front Genet, 2018
Chemical)disease,associa.ons,
Strong,evidence, Good,evidence, Limited,evidence,
Disease)disease,associa.ons,
Chemical,linkage,networks,
Disease, Disease, Score,
A, C, 0.50,
B, C, 0.38,
A, C, 0.21,
A,
C,
B,
A,
D,
B,
A,
C,E,
E,G,
F,
A,
C,
Chemical)disease,predic.on,
?,
Strong,evidence, Good,evidence, Limited,evidence,
Parkinson’s disease
Uterine cancer
Genito−urinary malformations
Cervical cancer
Vaginal cancer
Cholestasis
Altered time to sexual maturation
Reduced Fertility − male
Abnormal sperm
Adult−onset Leukemias
Reduced fertility − female
Hepatoma
Fetotoxicity
Immune suppression
Testicular toxicity
Renal cancer
Pancreatic cancer
Red blood cell damage
Stomach cancer
Acute non−lymphocytic
leukemia
Aplastic anemia
Menstrual disorders
Childhood Leukemias
Brain cancer − childhood
Salivary gland cancer
Brain cancer (adult)
Multiple myeloma
Thrombocytopenia
Thyroid cancer
Myelodysplastic syndrome
Nasopharyngeal cancer
Low birth weight
Pancreatitis
Porphyria
Sudden Infant Death Syndrome
Laryngeal cancer
Delayed growth
Fetal Alcohol Syndrome
Skeletal malformations Cardiac
congenital malformations
Esophageal cancer
Cerebrovascular disease
Pre−term delivery
Breast cancerDecreased vision
Cranio− Facial malformations
Hormonal changes
Systemic Lupus Erythematosus
Methemoglobinemia Raynaud’s phenomenon
Arrhythmias
Silicosis
Scleroderma
Angiosarcoma
Anemia
Diabetes − Type II
Autoimmune antibodies Acroosteolysis
Steatosis
Hepatitis
Cirrhosis
ADD/ADHD
Behavioral problems
Psychiatric disturbances
Berylliosis
Scrotal cancer
Skin cancer
Ovarian cancer
Mesothelioma
Pleural disease
Cognitive impairment
Hypertension
Cataracts
Gout
Melanoma (skin cancer)
Itai−itai disease
Chronic renal disease
Osteoporosis
Renal stones
Bronchitis − chronic
COPD
Cardiomyopathy
Asthma − irritant
Rhinitis − irritant
Bronchiolitis obliterans
Asbestosis Pneumoconiosis
Hard metal disease Congenital
malformations
Seizures
Decreased Coordination
Spasticity
Cerebral palsy
Hearing loss
Nasal septal perforation Nasal polyps
Myocardial infarction
Coronary artery disease
Minamata disease
Cancer
Cardiovascular
Connective tissue
Dermatological
Developmental
Ear, Nose, Throat
Endocrine
Gastrointestinal
Hematological
Immunological
Liver
Metabolic
Musculo-skeletal
Neurological
Ophthalmological
Psychiatric
Renal
Reproductive
Respiratory
20
Node size
Disease class
10
1
Development of network-based analysis tools to predict chemical-chemical
interactions to diseases and other biological outcomes
Pharmacological profiling: Network-based Analysis
Audouze K et al. PLoS One, 2014; Taboureau et al. ALTEX, 2017
Current collaborations with teams at P7
Alzheimer’s disease,DYRK1A, CBS (N. Janel)
NAT exploration with MD
simulation (F. Rodrigues-Lima)
Docking study into the CERT
transporter (C. Magnan)
RNAseq study on diabetes. Gut Microbiota peptides(C. Magnan)
Docking study: HSF2 (A. de Thonel and V. Lallemand-Mezger)
11-members including 7 permanents Permanent staff
PR. AC. Camproux
PR. O. Taboureau
CR. M. Petitjean
MCF. A. Badel
MCF. D. Flatters
MCF. L. Regad
Tech. C. Geneix
Post doc and Ph.D in 2018
ATER. V. Leroux
Ph.D N.. Cerisier
PhD. B. Boezio
PhD. P. Laville
Permanent staff
DR. P. Tuffery
MCF. G. Moroy
MCF. S. Murail
IG. J. Rey
IG. S. De Vries
Post doc and Ph.D in 2018Post doc: G. Postic
Team members
Thank you