EUGM 2014 - Céline Labbé (Institut National de la Santé et de la Recherche Médicale (INSERM)): iPPI-DB: A User-Friendly Web Application to Query a Database of Protein-Protein Interactions

iPPI-DB: A user-friendly web application to query a database of protein-protein

interactions inhibitors Céline Labbé

Inserm UMR-‐S 973 – MTi CDithem pla=orm

Paris

ChemAxon's 10th European User Group MeeHng May 20th-‐21st 2014

30 people

Sorbonne Paris Cité campus: 4 universi9es + 4 Ins9tutes in Paris 120,000 students 12,000 scien9sts in Life and Health Sciences 23 hospitals and 12,000 hospital beds

RPBS-‐MTI 958 64-‐Bits CPU core-‐linux

computer cluster 2x15 To data storage facility

MTi research unit & CDithem platform

Céline Labbé – EUGM Chemaxon – Budapest 2014 – 2/24

MTi at University Paris Diderot

www.mH.univ-‐paris-‐diderot.fr

www.CDithem.com


Ø  High number of protein-‐protein interacHon (PPI)

Ø  Involvement in various diseases (eg cancer)

Ø  Importance of finding modulators of PPI for therapeuHc intervenHon

Why studying the PPI ?




Ø  Importance of finding modulators of PPI for therapeuHc intervenHon Ø  Usually use of High Throughput Screening (HTS)

Ø  Inadequacy of commercial chemical libraries

Ø  MisconcepHon of the PPI chemical space





Ø  Importance of finding modulators of PPI for therapeuHc intervenHon

Ø  Strategy: learning from successful examples

•  data collecHons on PPI and inhibitors of PPI (iPPI)

•  characterizaHon of the PPI chemical space

•  creaHon of PPI focused chemical libraries

Ø  Usually use of High Throughput Screening (HTS)

Ø  Inadequacy of commercial chemical libraries

Ø  MisconcepHon of the PPI chemical space



Existing databases

Targets Manual data curaHon Contents

All types

Chemical structures

•  Chemical structure •  Physicochemical characterisHcs •  Pharmacological data

No of PPI target

- 

- 

ü 

ü 

X-‐Ray only

No of compounds

1,300,000

71

7,000

1,650

44

50

31

NA

•  Chemical structure •  Pharmacological data

•  Cocrystallized structure

•  Chemical structure •  Physicochemical characterisHcs •  Pharmacological data

PPI only

PPI only

PPI only

ü 

ü 

ü 


iPPI-‐DB: what type of informations ? Compound Biblio

PPI / Proteins Test AcHvity

Ø  Molecular descriptors

Ø  Compounds names

Ø  External references

-‐> AlogP, Molecular Weight, Fsp3…

-‐> IUPAC, brand name

Ø  Type of test

Ø Test name

Ø  Type of acHvity

Ø  AcHvity value

-‐> IC50, EC50, Kd, Ki

Ø  Pubmed ID or Wipo ID

-‐> arHcle, patent

Ø  Title

Ø  Journal

Ø  Year of publicaHon

Ø  Pair of protein names

Ø  Uniprot number

-‐> ELISA, fluorescence polarizaHon…

-‐> Biochemical or cellular test


Ø  Source •  Liierature (PubMed), world patents •  Manually curated by a medicinal chemist

Ø  Criteria •  AcHvity : IC50, Ki, Kd, EC50 < 30 μM •  Absence of reacHve or promiscuous-‐associated chemical funcHons •  Rule out pepHdes (Absence of 3 conHnuous pepHde bonds) •  Rule out macrocycles •  Degree of validaHon of the target •  Clarity of the experimental data on binding

Ø  Stats

iPPI-‐DB: criteria


iPPI-‐DB: some numbers

695

527

277

349

122

73

303

40

24

11

8

5

1

460

326

277

268

119

73

46

40

17

10

8

5

1

0 200 400 600 800

MDM2-‐like/p53

BCL2-‐like/BAX

LFA/ICAM

XIAP/Smac

CD4/gp120

CD80/CD28

Bromodomain/histone

Beta-‐catenin/TCF-‐4

IL2/IL2R

E2/E1

Myc/max

LEDGF/IN

ZipA/osZ

Number of compounds and binding data per PPI target

No. of unique compounds

No. of binding data

0 1 2 3 4 5 6 7 8 9

Number of iPPI in clinical trials per PPI target (from MDDR data)

Phase II Phase I Preclinic


iPPI-‐DB: some numbers

2

40

245

3 7 1 1 24

95

15 18 3

31 14

4

74

0

50

100

150

200

250

300

Cellular Tests

pXC50 ≥ 8 7 > pXC50 ≥ 6 8 > pXC50 ≥ 7 6 > pXC50

pXC50 = -‐ log(XC50x10-‐6)

iPPI-‐DB: the web application


Ø  Provide the PPI community with a user-‐friendly online interface

Ø  Facilitate the access to the right informaHon

•  user defined criteria

•  cross referencing the annotated data

Ø  Help to prioriHze the selecHon of privileged chemotypes and physicochemical properHes

Ø  Can be accessed by anyone at www.ippidb.cdithem.fr

Search by pharmacological criteria -‐ Query



Ø  Only the selecHon of a PPI target is mandatory




Ø  Only the test and acHvity type available for the selected PPI target are proposed in the drop-‐down menus





Ø  pXC50 : eg

•  pIC50 = -‐ log(IC50x10-‐6)

•  pKd = -‐ log(Kdx10-‐6) or





Ø  pXC50 : eg

•  pIC50 = -‐ log(IC50x10-‐6)

•  pKd = -‐ log(Kdx10-‐6) or





Ø  pXC50 : eg

•  pIC50 = -‐ log(IC50x10-‐6)

•  pKd = -‐ log(Kdx10-‐6)

Ø  Fsp3 =

or

No of Carbon sp3

No Total of Carbon



Search by pharmacological criteria -‐ Results











The compound ID Card -‐ Summary











Ø  3 / 75 filter : rule for in vivo toxicity from Pfizer

The compound ID Card -‐ Physicochemistry



all iPPI descriptors' values should be ideally within the blue area




all iPPI descriptors' values should be ideally within the blue area

PCA : Principal Component Analysis



LE : Ligand Efficiency

1.37 × pXC50

No Heavy Atoms LE =

LLE : Lipophilic Efficiency

LLE = pXC50 − AlogP

The compound ID Card -‐ Pharmacology


Ø  Annotated from MDDR data (march 2012)

Ø  FCFP: FuncHonal-‐Class Fingerprints

Ø  Similarity : Tanimoto index between two fingerprints

really dissimilar molecules 0

1

same molecules

The compound ID Card – Drug similarity

More information ?


PMID: 23688585

Search for drug candidates



Search for drug candidates


Search for drug candidates -‐ Query


Search for drug candidates -‐ Query


Search for drug candidates -‐ Results


Search for drug candidates -‐ Results

Towards the next version of the web app


Log in

Get your results !


Log in

Get your results !



Log in

Get your results !


Search by chemical similarity -‐ Query


Ø  Copy / paste a SMILES

Import a file

Sketch your molecule or

or



Import a file

Sketch your molecule

Ø  Example from :

or

or

Nutlin-‐1 (acHvity on MDM2)




Import a file

Sketch your molecule

Ø  Example from :

or

or

Nutlin-‐1 (acHvity on MDM2)



Search by chemical similarity -‐ Results


Search by chemical similarity -‐ Results

Conclusions


Ø  a database

ü  manually curated by experts

ü  containing :

-  chemical structures

-  physicochemical properHes

-  pharmacological data

iPPI-‐DB

Conclusions


Ø  a database


ü  containing :




Ø  a user-‐friendly web applicaHon

ü  search by pharmacological criteria

ü  search by chemical similarity

ü  cross-‐referencing the annotated data

ü  intuiHve visualizing tools

iPPI-‐DB

Conclusions


Ø  a database


ü  containing :




Ø  a user-‐friendly web applicaHon

ü  search by pharmacological criteria

ü  search by chemical similarity

ü  cross referencing the annotated data

ü  intuiHve visualizing tools

iPPI-‐DB

Assist chemists, biologists and clinicians to design more raHonaly the next generaHon of PPI modulators

Acknowledgement


Ø  Mélaine Kuenemann (PhD Student)

Ø  David Lagorce (Engineer)

Ø  Maria Miteva (Team Leader)

Ø  Olivier Sperandio (Senior Researcher)

Ø  Bruno Villoutreix (Unit director)

Ø  Judith Elkaim (former member)

Ø  Guillaume Laconde (former member)

Ø  Pauline Raballand (former member)

Ø  Benoît Déprez (Experimentalist)

Ø  Jean-‐Luc Poyet (Experimentalist)

Software

EUGM 2014 - Céline Labbé (Institut National de la Santé et de la Recherche Médicale (INSERM)): iPPI-DB: A User-Friendly Web Application to Query a Database of Protein-Protein Interactions