Laboratoire de Chemoinformatique

2013

infochimie — 2013-04-15T14:03:10Z

ECO Winter School in Kalmar (Sweden) between the 25th February and the 1st March 2013.
Formation Continue : Initiation à la Chemoinformatique à Strasborug du 13 Mai 2013 au 19 Mai 2013/Short Courses in Chemoinformatics in Strasbourg between the 13th May 2013 and the 19th May 2013.

Short Courses in Chemoinformatics

infochimie — 2013-04-15T13:54:48Z

Short Courses in Chemoinformatics

from 13th May 2013 to 19th May 2013

Université de Strasbourg, Faculté de Chimie, Strasbourg, FRANCE

Reference : CBT09-0684A

Duration

5 days, 13-19 May 2013

For whom ?

Chemists (Bachelor's degree or better), technicians having experience in database management, modelisation software, willing to broaden their skills. Chemoinformatics is a new scientific area appeared a few years ago. It relates to (i) storage, management and search of chemical data (databases), (ii) the development of predictive models connecting the chemical structure of molecules with their physicochemical and biological properties, and, (iii) design in silico of new compounds having desired properties. The chemoinformatics approaches are very much used in phamaceutical companies to carry out a virtual screening, to optimize a lead, for predictions of ADME/Tox properties.

Requirements

Basic informatics skills.

Language

Lectures will be delivered in English or in French depending on the audience.

Location

Université de Strasbourg

Faculté de Chimie

1, rue Blaise Pascal

67070 STRASBOURG

The courses will take place in the computer class, equipped 21 PC double-boot Windows/LINUX, 4 work stations SGI, a printer and a video projector.

Program

Computer representation of chemical structures and electronic exchange formats
Database management systems in chemistry.
Descriptors, pharmacophores, chemical space
Similarity, diversity, clustering : design and analyzes of chemical libraries.
SAR classification models (Decision trees, SVM, Networks of Neurons) and QSAR regression models (MLR, PLS).
Validation and applicability domains of QSAR models.
Docking. Virtual screening, hit selection. Lead optimization.

Softwares used in the lectures

ChemAxon, ISIDA, WEKA, MOE, LigandScout, FlexX.

Registration fees

1.810 €

These fees include teaching and lunch

Informations and registration

DEPULP

Sandra GRISINELLI

Tel. 03 68 85 49 98

Except on Wednesday

Fax : 03 68 85 49 29

Director

Alexandre Varnek, Professor at Strasbourg University.

e-mail : varnek@unistra.fr

Formation Continue : Initiation à la Chemoinformatique

infochimie — 2013-04-15T13:45:31Z

Initiation à la Chemoinformatique

du 13 Mai 2013 au 17 Mai 2013

Université de Strasbourg, Faculté de Chimie, Strasbourg, FRANCE

Référence : CBT09-0684A

Présentation

Durée : 5 jours

Calendrier : du 13 Mai 2013 à 9h au 17 Mai 2013 à 16 h

Personnes concernées

Chimistes (Bac+3 ou plus), techniciens supérieurs (DUT) ayant une expérience en gestion des bases de données, logiciels de modélisation, souhaitant élargir leur domaine de compétence. La Chemoinformatique est un nouveau domaine apparu il y a quelques années. Il concerne (i) le stockage, la gestion et la recherche de données chimiques (bases de données), (ii) le développement de modèles prédictifs reliant la structure chimique de molécules avec leurs propriétés physico-chimiques et biologiques, et, (iii) le design in silico de nouveaux composés possédant les propriétés désirées. Les approches Chemoinformatique sont trés utilisées dans l'industrie pharmaceutique pour effectuer un criblage virtuel, optimiser un lead ainsi que pour des prédictions de propriétés ADME/Tox.

Pré-requis

Connaissance de base en informatique.

Langue

Les cours seront délivrés en Anglais ou Français en fonction de l'audience.

Lieu

Université de Strasbourg

Faculté de Chimie

1, rue Blaise Pascal

67070 STRASBOURG

L'enseignement se déroulera au sein de la Faculté de Chimie, dans une salle réservée à cette formation, équipée de 21 PC en double boot Windows/LINUX, de 4 stations de travail SGI, d'une imprimante et d'un vidéo projecteur.

Programme

Représentation de structures chimiques par ordinateur et formats électroniques d'échange
Systèmes de gestion des bases de données en chimie.
Descripteurs, pharmacophores, espace chimique
Similarité, diversité, clustering : conception et analyse de chimiothéque.
Modèles QSAR de classification (Arbres de Décision, SVM, Réseaux de Neurones) et de régression (MLR, PLS)
Validation et domaine d'applicabilité des modèles QSAR
Docking. Criblage virtuel, sélection des hits. Optimisation d'un lead.

Logiciels utilisés dans les cours

ChemAxon, ISIDA, WEKA, MOE, LigandScout, FlexX.

Frais de participation

1.810 €

Cette participation inclu l'enseignement et les repas de midi

Renseignements et inscriptions

DEPULP

Sandra GRISINELLI

Tél. 03 68 85 49 98

Sauf le mercredi

Fax : 03 68 85 49 29

s.grisinelli@unistra.fr

Responsable scientifique

Alexandre Varnek, Professeur à la Faculté de Chimie.

Courriel : varnek@unistra.fr

ISIDA/ModelAnalyzer

infochimie — 2013-02-23T22:00:53Z

ModelAnalyzerC

This software uses a text file containing predictions of a classification models. The text file shall be organized as follows :

if a line starts with a "#", it is a command or a comment if it is not interpreted
possible commands are :
- #SDF sdfile.sdf
- #Classes : the number of classes of the classifications problem. For a binary classification, this number is 2.
- #Predictions : the number of column containing classification outputs. If only the file concerns only one model, the number value is 1.
- #Weights ... : This line contains an array of values, at least one per classification outputs. These values are used to weight the vote of classification models for a vote.
A data lines shall be structures as follows :
- First column is the ID of the sample
- Second column, the reference/experimental value
- Next #Predictions columns are the actual class assignments by the model
- Next #Predictions columns are real values interpreted as a confidence score of the class assignment. One value per model.

If an SDF is provided, the compounds are supposed to follow the same order in the SDF file as referenced by the IDs. The ID of the first molecule shall be 1 and so on.

The software supports files ".out" which are expected to be Weka outputs containing class assignment for each instances in CSV format.

Mac version (mountain lion) : download

Windows : download

ModelAnalyzerR

This software uses a text file containing predictions of a regression models. The text file shall be organized as follows :

if a line starts with a "#", it is a command or a comment if it is not interpreted
possible commands are :
- #SDF sdfile.sdf
- #Predictions : the number of column containing classification outputs. If only the file concerns only one model, the number value is 1.
- #Weights ... : This line contains an array of values, at least one per classification outputs. These values are used to weight the vote of classification models for a vote.
A data lines shall be structures as follows :
- First column is the ID of the sample
- Second column, the reference/experimental value
- Next #Predictions columns are the actual estimates of the model
- Next #Predictions columns are real values interpreted as a confidence score of the estimates. One value per model.

If an SDF is provided, the compounds are supposed to follow the same order in the SDF file as referenced by the IDs. The ID of the first molecule shall be 1 and so on.

The software supports files ".out" which are expected to be Weka outputs containing estimates for each instances in CSV format.

Mac version (mountain lion) : download

Windows : download

ECO Winter School, Kalmar (Sweden)

infochimie — 2013-02-21T18:59:39Z

This page is dedicated to the tutorial on ensemble modeling presented at the ECO Winter School in Kalmar (Sweden) between the 25th February and the 1st March 2013.

Programme


Agenda ECO Winter School

The tutorial was designed initially for the Chemoinformatics Summer School in 2010 : CS3 2010

Instructions :


Tutorial on Ensemble Modeling in Chemoinformatics

Datasets

Pharmacophore

infochimie — 2012-06-07T17:26:46Z

Tutors

Sharon Bryan (Inte:Ligand), Jean-Christophe Mozziconacci (Schrödinger)

Software

Inte:Ligand
- LigandScout (http://www.inteligand.com/ligandscout/)
Schrödinger
- PhaseShape (http://www.schrodinger.com/products...)
- Phase (http://www.schrodinger.com/products...)
- Maestro (http://www.schrodinger.com/download...)

Maestro can be downloaded for free for all academic users.

Datasets

The docking/pharmacophore dataset consists of 77 compounds : 11 active compounds and 66 decoys randomly selected from the QSAR classiﬁcation dataset. Structures are provided with 3D coordinates in mol2 format in the file named P29274_docking-random.mol2. Of note, one of the active compounds and 8 of the decoys have an undefined chiral center, hence the two possible stereoisomers were included in the dataset. The named P29274_docking-random.mol2 thus contains 86 structures. The structure name allows one to distinguish active compounds ("chembl" prefix) from decoys ("pubchemdecoy" prefix).

In order to be able to compare individual results and prepared benchmarks, we advise you to use the PDB entry 3EML, which you can download from the PDB (http://www.rcsb.org).

Download : LigandScout dataset

Download : LigandScout supplementary files. A2A virtual screening results using LigandScout on 1 million commercially available compounds.

Download PhaseShape supplementary files

Instructions

LigandScout tutorial (to be used with this dataset)
Training videos are available for Glide and Maestro here
PhaseShape tutorial

Docking

infochimie — 2012-06-07T16:49:02Z

Tutors

Marcus Gastreisch (BioSoveIt), David Rinaldo (Schrödinger), Esther Kellenberger (Univ. Strasbourg)

Software

BioSolveIT
- LeadIT (http://www.biosolveit.de/LeadIT/ind...)
Schrödinger
- Glide (http://www.schrodinger.com/products...)
- Maestro (http://www.schrodinger.com/download...)

Maestro can be downloaded for free for all academic users.

Datasets

LeadIT tutorial
Training videos are available for Glide and Maestro here
Glide tutorial

QSAR

infochimie — 2012-06-07T16:47:10Z

Tutors

Tetko (eADMET), Gilles Marcou (Univ.Strasbourg)

Software

The following software will be used for the tutorials :

OCHEM (http://ochem.eu/)
WEKA (http://www.cs.waikato.ac.nz/ml/weka/)

_ _

Organization of the session

During the session both Weka and OChem will be briefly presented. Participants shall be free to explore the software to find their own solutions with the help of the tutors.

You can find short instructions and examples of Weka input files for the A2A problems at the end of this page.

Datasets

Two studies are proposed : (1) the prediction of the affinity of each ligand for A2AR by regression and (2) the binary classification of all ligands into classes of active and inactive compounds. Experimental affinity values are expressed as pKi, pKD, pEC50 or pIC50, but for the sake of simplicity, only the unit pKi will be used later in this document.

Two datasets are provided for QSAR studies : one for regression and the other for classiﬁcation. Structures are provided with 2D coordinates in MDL SDF format.

Dataset for regression :
A2AR ligands and their pKi values were collected in IupharDB, ChEMBL and PubChem BioAssay. If several values were found for a given ligand, the reported value is the median of the collected values if the range of values did not exceed 2 log units.
The dataset contains 766 compounds : 384 in a training set and 383 in a test set.
The affinity value of each compound is stored in the SDF field pKi, in log units. The training set and the test set files are named respectively train.sdf and test.sdf and are located in the Regression folder.

Download the dataset

A2A Regression problem : ISIDA descriptors in Weka (ARFF) format (use this file with the instructions).


ISIDA_A2AR: ISIDA descriptors for the A2A regression problem

Download different types of ISIDA descriptors for regression

A2A Regression problem : MOE2D descriptors in Weka (ARFF) format.


MOE2D_A2AR: MOE2D descriptors for the A2A regression problem

Dataset for classification :
Starting from the regression dataset, a compound was classified as active if pKi>5. Decoys were randomly chosen from PubChem and have no reported affinity for A2A. The number of decoys was set so that the ratio between active and inactive compound is about 15%. The classiﬁcation dataset contains 4948 compounds (756 actives ; 4192 inactives), 2430 in the training set (381 actives ; 2049 inactives) and 2441 compounds in the test set (364 actives ; 2077 inactives). Random picking in the test set yielded a smaller test set made of 77 compounds (11 actives ; 66 inactives) that will also be tested in docking and pharmacophore methods.
Activity is reported in the SDF field Activity : 1 for an active compound and 0 for an inactive compound. The training set and the test set files are named respectively train.sdf and test.sdf. They are located in the Classification folder.

Download the dataset

A2A Classification problem : ISIDA descriptors in Weka (ARFF) format (use this file with the instructions).


ISIDA_A2AC: A2A Classification ISIDA descriptors

A2A Classification problem : MOE2D descriptors in Weka (ARFF) format. Part 1 contains the training and Part 2 contains the test sets.


MOE2D_A2AC-1: MOE2D descriptors for the A2A classification problem -Part 1.


MOE2D_A2AC-2: MOE2D descriptors for the A2A classification problem -Part 2.

Download different types of ISIDA descriptors for classification

Note that the name of the compounds does not necessarily refer to an existing entry into one of the databases from which the datasets were extracted.

Instructions for Weka and input files on A2A

Short description of the software and examples of modelling sessions using Weka on A2A problems :


Short_Instructions_Weka: Short instructions on using Weka for the tutorials.

Files to use with the instructions

A2A Classification problem : ISIDA descriptors in Weka (ARFF) format


ISIDA_A2AC: A2A Classification ISIDA descriptors

A2A Regression problem : ISIDA descriptors in Weka (ARFF) format


ISIDA_A2AR: ISIDA descriptors for the A2A regression problem

Download different types of ISIDA descriptors for regression

Download different types of ISIDA descriptors for classification

A2A Classification problem : MOE2D descriptors in Weka (ARFF) format. Part 1 contains the training and Part 2 contains the test sets.


MOE2D_A2AC-1: MOE2D descriptors for the A2A classification problem -Part 1.


MOE2D_A2AC-2: MOE2D descriptors for the A2A classification problem -Part 2.

A2A Regression problem : MOE2D descriptors in Weka (ARFF) format.


MOE2D_A2AR: MOE2D descriptors for the A2A regression problem

Examples of WEKA modelling

Tutorials

infochimie — 2012-06-07T16:29:47Z

Organization

Three tutorials will be organized for the participants of the Summer School : (i) QSAR, (ii) pharmacophore models and (iii) docking. They will take place in 3 computer classes at the 4th floor of the Faculty of Chemistry building (1, rue B. Pascal). Each class accommodates up to 25 people. So, all participants will be divided into 3 groups each following its schedule :

Group 1 : QSAR, pharmacophore+docking 1 (with BiosolveIT and Inte:Ligand tools), pharmacophore+docking 2 (with Schrödinger tools)
Group 2 : pharmacophore+docking 1 (with BiosolveIT and Inte:Ligand tools), QSAR, pharmacophore+docking 2 (with Schrödinger tools)
Group 3 : pharmacophore+docking 2 (with Schrödinger tools), pharmacophore+docking 1 (with BiosolveIT and Inte:Ligand tools), QSAR.

Three parallel sessions are scheduled :

Session 1 : Tuesday 26 June from 14:00 to 16:00 ;
Session 2 : Wednesday 27 June from 14:00 to 16:00,
Session 3 : Wednesday 27 June from 16:30 to 18:30,

Thus each participant is supposed to attend all three tutorials. A briefing resuming the tutorial sessions will be organized on Thursday 27 June in the conference hall in the CDE building.
The participants are encouraged to download the data and software tools for the tutorials and to work on the exercises before the school.

Data

The tutorials focus on the type A2A of human adenosine receptor (A2AR). A2AR is a member of purinergic class of G-Protein Coupled Receptors. It is one of the main targets of caffeine, and also of some drugs used for coronary vasodilation (agonist compounds) or for the treatment of Parkinson's disease (antogonist compounds). The three-dimensional structure of A2AR in complex with the antagonist ZM241385 was solved in 2008 (Science. 2008, 322 : 1211–1217). Since then, ten other structures of the receptor were solved in different states of activation and in complex with different ligands.

The tutorials aim at predicting the binding of ligands to A2AR (whatever their function, agonist or antagonist) using three different approaches : QSAR, pharmacophore models and docking.

Poster Session

infochimie — 2012-06-07T15:54:53Z

Strasbourg Summer School - Strasbourg, France, 25-29 June 2012

Tuesday 26 June - Poster session *

P1	Elena Mokshina	“Quasi-mixture”-based QSPR study of critical parameters of pure organic compounds
P2	Kamel Mansouri	QSAR STUDY ON READY BIODEGRADABILITY OF CHEMICALS
P3	Nicholas Firth	Plane of Best Fit : A Novel Method to Characterize the Three-Dimensionality of Molecules
P4	Charfedinne Ayed	Exploring subtle molecular interactions involved between aroma compounds and food matrices/mediums using QSPR approach
P5	David Hoksza	Molecular morphing – novel method for chemical space exploration
P6	Satoshi Niijima	Dissecting kinase profiling data to predict activity and understand cross-reactivity of kinase inhibitors
P7	Eugene Muratov	QSARome of GPCRome
P8	Daria Tsareva	Applying Machine Learning to Classification of TRPV1 Antagonists
P9	Andreas Jurik	Combining Ensemble Docking and Consensus Scoring to Elucidate the Binding Mode of Tiagabine Analogs in hGAT-1
P10	Florian Koelling	X-PLORATION SPACE ! plif driven parallel screening (PLIDripas) in progress …
P11	Amir Seddik	Common Scaffold Clustering as a Versatile Tool for Prioritizing Docking Poses of Amphetamine Derivatives
P12	Claire Minoletti	Chemoinformatics tools for CYP induction assessment
P13	Matteo Cassotti	Comparison of variable selection methods
P14	Sylvain Lozano	Prediction of the CYP3A4 inhibition potential of new chemicals : a ligand-based approach.
P15	Florent Chevillard	Growing Specific Ligands for β2AR and HH1R from a Common Fragment-Sized Scaffold
P16	J.B. Brown	Virtual screening via kernel methods incoporating charge, chirality, and rigidity
P17	Latifeh Navidpour	Flavonoids as Human a-amylase inhibitors : QSAR analysis
P18	Azadeh Ebrahim-Habibi	Non carbohydrate-based small molecule inhibitors of mammalian alpha-amylase : toward rational design of novel compounds
P19	Robert Körner	In silico pKa prediction
P20	Véronique Lancelot	Homology modeling of nicotinic acetylcholine receptors
P21	Leslie REGAD	Characterization of inhibitor binding sites of different sub-families of Serine Proteases
P22	Sabrina Wollenhaupt	Comparison of fingerprint- and MCS-based (INSARA) Structure-Activity-Relationship networks
P23	Yuye HE	Development of Novel Methods for Evaluating Prediction Models
P24	Akira Shiraishi	Chemical genomic analysis of cross-reactivity in endogenous GPCR-mediated pathways
P25	Salwa Soliman	Ligand- and structure-based modeling of human Aryl Sulfotransferase 1A1 activity
P26	M. Stepniewski	role of PEG in drug delivery STUDIED BY molecular dynamics simulation
P27	Yuezhou Zhang	Structure-based screening and discovery of inhibitors of the human macrodomain protein 1
P28	Guillaume Poezevara	Emerging Patterns as Structural Alerts for Computational Toxicology
P29	Ahmed Abdelaziz	Using Toxcast™ HTS assays as a biologically derived descriptors in QSAR
P30	Jamil Al-Asri	Pharmacophore creation for the discovery of new alpha-amylase inhibitors
P31	Fabienne Dulin	New insights into the mode of action of AChE inhibitors pesticides in honeybees : molecular modeling studies
P32	Jade Fogha	Analysis of MCL-1 and BH3-only proteins. A model for drugs discovery
P33	Alban Lepailleur	Towards new ligands with pro-cognitive properties. Virtual screening and polypharmacology.
P34	Véronique Hamon	Inhibition of Protein-Protein Interfaces : New tools to filter PPI chemical space through Support Vector Machine approaches
P35	David A. Demeter	Improving the Impact of Compound Acquisition : Comparison of Biology-based Models with Physiochemical Rule-based Models for Hit Generation
P36	Jeremy DESAPHY	Protein/ligand interactions : From a simple representation to 3D alignment using graph matching algorithm
P37	Jamel Meslamani	Profiler : A hybrid pipeline for in-silico biological profiling of small molecules
P38	Jamel Meslamani	Utility Assessment of Protein−Ligand-Based Pharmacophores in Computational Ligand Profiling
P39	Crina-Maria Ionescu	Quantum Mechanical Evaluation of Charge Transfer inside Proteins
P40	Ondřej Skřehota	QSPR Designer – Automate the QSAR modeling process and employ own descriptors
P41	Stanislav Geidl	Predicting pKa values of substituted phenols by QSPR models which employ EEM atomic charges
P42	Ctibor Škuta	Data mining of HTS data of steroid compounds and discovery of relations between their chemical structure and biological activity
P43	Natalia Kireeva	Design of "Optimal" Descriptor Spaces for Data Visualization and Modeling
P44	Fiorella Ruggiu	ISIDA Property-Labelled Fragment Descriptors
P45	Héléna A. Gaspar	Generative Topographic Mapping : a universal tool for chemical space visualization, activity prediction and databases comparison
P46	Christophe Muller	Prediction of oxidations sites for human CYP3A4 substrates using Condensed Graph of Reactions.
P47	Tetiana Khristova	Computer-aided design of new inhibitors of platelets aggregation : RGD-peptidomimetics
P48	Laurent Hoffer	Sampling, Docking & Fragment-based Drug Design
P49	Aurélie De Luca	Predictive models for acido-basic properties of molecules and reactions using new reactivity descriptors
P50	Fanny Bonachera	Using self-organizing maps to accelerate similarity search
P51	Ioana Oprisiu	Modeling of non-additive mixtures properties using On-line CHEmical database and Modeling Environment (OCHEM)
P52	Véronique Lancelot	Master « In silico Drug Design », Association and Website

* Only the name of the first author is mentioned in the list