E-Chemistry and Web 2. 0 Marlon Pierce mpierce@cs. indiana. edu Community Grids Lab Indiana University 1
One Talk, Two Projects £ NIH funded Chemical Informatics and Cyberinfrastructure Collaboratory (CICC) @ IU. ¤ Geoffrey Fox ¤ Gary Wiggins ¤ Rajarshi Guha ¤ David Wild ¤ Mookie Baik ¤ Kevin Gilbert ¤ And others £ Proposed Microsoft. Funded Project: EChemistry ¤ Carl Lagoze (Cornell), ¤ Lee Giles (PSU), ¤ Steve Bryant (NIH), ¤ Jeremy Frey (Soton), ¤ Peter Murray-Rust (Cambridge), ¤ Herbert Van de Sompel (Los Alamos), ¤ Geoffrey Fox (Indiana) ¤ And others 2
CICC Infrastructure Vision £ Chemical Informatics: drug discovery and other academic chemistry, pharmacology, and bioinformatics research will be aided by powerful, modern, open, information technology. ¤ NIH Pub. Chem and Pub. Med provide unprecedented open, free data and information. ¤ We need a corresponding open service architecture (i. e. avoid stove-piped applications) ¤ CICC set up as distributed cyberinfrastructure in e. Science model £ Web clients (user interfaces) to distributed databases, results of high throughput screening instruments, results of computational chemical simulations and other analyses. ¤ Composed of clients to open service APIs (mash-ups) ¤ Aggregated into portals £ Web services manipulate this data and are combined into workflows. £ So our main agenda items: create interesting databases and build lots of Web services and clients. 3
CICC Databases £Most of our databases aim to add value to Pub. Chem or link into Pub. Chem ¤ 1 D (SMILES) and 2 D structures £ 3 D structures (MMFF 94) ¤Searchable by CID, SMARTS, 3 D similarity £Docked ligands (FRED, Autodock) ¤ 906 K drug-like compounds into 7 ligands ¤Will eventually cover ~2000 targets £Philosophy: we have big computers, so let’s calculate everything ahead of time and put the results in a DB.
Building Up the Infrastructure £ Our SOA philosophy: use standard Web services. ¤Mostly stateless ¤Some cluster, HPC work needed but these populate databases £ Services are aggregate-able into different workflows. ¤Taverna, Pipeline Pilot, … £ You can also build lots of Web clients. £ See http: //www. chembiogrid. org/wiki/index. php/CICC_ Web_Resources for links and details. £ Not so far from Web 2. 0…. 5
Sample Services 6
7
8
9
10
Web Client Interfaces 11
More Clients… 12
More Clients… 13
Example: Pub. Dock £ Database of approximately 1 million Pub. Chem structures (the most drug-like) docked into proteins taken from the PDB £ Available as a web service, so structures can be accessed in your own programs, or using workflow tools like Pipeline Polit £ Several interfaces developed, including one based on Chimera (right) which integrates the database with the PDB to allow browsing of compounds in different targets, or different compounds in the same target £ Can be used as a tool to help understand molecular basis of activity in cellular or image based assays 14
Example: R Statistics applied to Pub. Chem data £ By exposing the R statistical package, and the Chemistry Development Kit (CDK) toolkit as web services and integrating them with Pub. Chem, we can quickly and easily perform statistical analysis and virtual screening of Pub. Chem assay data £ Predictive models for particular screens are exposed as web services, and can be used either as simple web tools or integrated into other applications £ Example uses DTP Tumor Cell Line screens - a predictive model using Random Forests in R makes predictions of probability of activity across multiple cell lines. 15
A protein implicated in tumor growth with known ligand is selected (in this case HSP 90 taken from the PDB 1 Y 4 complex) The screening data from a cellular HTS assay is similarity searched for compounds with similar 2 D structures to the ligand. Similar structures to the ligand can be browsed using client portlets. Example assay screening workflow: finding cell-protein relationships Similar structures are filtered for drugability, are converted to 3 D, and are automatically passed to the Open. Eye FRED docking program for docking into the target protein. Docking results and activity patterns fed into R services for building of activity models and correlations Least Squares Regression Random Forests Neural Nets Once docking is complete, the user visualizes the highscoring docked structures in a portlet using the JMOL applet. 16
Relevance to Web 2. 0 £ Some Web 2. 0 Key Features ¤REST Services ¤Use of RSS/Atom feeds ¤Client interfaces are “mashups” ¤Gadgets, widgets for portals aggregate clients £ So… ¤We provide RSS as an alternative WS format. ¤We have experimented with RSS feeds, using Yahoo Pipes to manipulate multiple feeds. ¤CICC Web interfaces can be easily wrapped as universal gadgets in i. Google, Netvibes. ¥Alternative to classic science gateways. 17
RSS Feeds/REST Services £Provide access to DB's via RSS feeds £Feeds include 2 D/3 D structures in CML £Viewable in Bioclipse, Jmol as well as Sage etc. £Two feeds currently available ¤Syn. Search – get structures based on full or partial chemical names ¤Dock. Search – get best N structures for a target £Really hampered by size of DB and Postgres performance.
Tools and mashups based on web service infrastructure http: //www. chembiogrid. org/projects/proj_tools. html 19
Mining information from journal articles £ Until now Sci. Finder / CAS only chemistry-aware portal into journal information £ We can access full text of journal articles online (with subscription) £ ACS does not make full text available … but there are ways round that! £ RSC is now marking up with SMILES and GO/Goldbook terms! ¤ www. projectprospect. org £ Having SMILES or In. Ch. I means that we can build a similarity/structure searchable database of papers: e. g. “find me all the papers published since 2000 which contain a structure with >90% similarity to this one” £ In the absence of full text, we can at least use the abstract 20
Text Mining: OSCAR £ A tool for shallow, chemistry-specific natural language parsing of chemical documents (e. g. journal articles). £ It identifies (or attempts to identify): ¤ Chemical names: singular nouns, plurals, verbs etc. , also formulae and acronyms. ¤ Chemical data: Spectra, melting/boiling point, yield etc. in experimental sections. ¤ Other entities: Things like N(5)-C(3) and so on. £ Part of the larger Sci. Borg effort ¤ See http: //www. cl. cam. ac. uk/~aac 10/escience/sciborg. html) £ http: //wwmm. ch. cam. ac. uk/wikis/wwmm/index. php/Os car 3 21
Create a database containing the text of all recent Pub. Med abstracts (2006 -2007 = ~500, 000) Use OSCAR to extract all of the chemical names referred to in the abstracts and covert to SMILES DATABASE SERVICE + DOCKING SERVICE Convert molecules to 3 D and dock into a protein of interest Visualize top docked molecules in a Googlelike interface Mash-Up: What published compounds might bind to this protein? 22
E-Chemistry and Digital Libraries We can’t wait to get started…. 23
E-Chemistry and Digital Libraries £ Key problem with our SOA-based e-Science is information management. ¤Where is the service that I need? ¤What does it do? £ We may consider our data-centric services to be digital libraries. £ Data is diverse ¤Documents ¤Not just computational information like structures. £ Another point of view: how can I link together publications, results, workflows, etc? ¤That is, I need to manage digital documents. 24
Digital Libraries £ Open Archives Initiative Object Reuse and Exchange Project (OAI-ORE) £ Developing standardized, interoperable, and machinereadable mechanisms to express information about compound information objects on the web. £ Graph-based representations of connected digital objects. £ Objects may be encoded in (for example) RDF or XML, £ Retrievable via repositories with REST service interfaces (c. f. Atom Publishing Protocal) ¤ Obtain, harvest, and register 25
26
27
Challenges for E-Chemistry £ Can digital library principals be applied to data as well as documents? ¤Can you link your workflow to your conference paper? £ Can we engineer a publishing framework and message formats around Web 2. 0 principals? ¤REST, Atom Publishing Protocol, Atom Syndication Format, JSON, Microformats £ Can we do this securely? ¤Access control, provenance, identify federation are key problems. 28
29
More Information £Project Web Site: www. chembiogrid. org £Project Wiki: www. chembiogrid. org/wiki £Contact me: mpierce@cs. indiana. edu 30
31
CICC Chemical Informatics and Cyberinfrastucture Collaboratory Funded by the National Institutes of Health www. chembiogrid. org CICC Combines Grid Computing with Chemical Informatics Large Scale Computing Challenges Chemical Informatics is non-traditional area of high performance computing, but many new, challenging problems may be investigated. NIH Pub. Med Data. Base Chemical informatics text analysis programs can process 100, 000’s of abstracts of online journal articles to extract chemical signatures of potential drugs. OSCAR Text Analysis Initial 3 D Structure Calculation Molecular Mechanics Calculations Cluster Grouping Toxicity Filtering Science and Cyberinfrastructure CICC is an NIH funded project to support chemical informatics needs of High Throughput Cancer Screening Centers. The NIH is creating a data deluge of publicly available data on potential new drugs. . Docking OSCAR-mined molecular signatures can be clustered, filtered for toxicity, and docked onto larger proteins. These are classic “pleasingly parallel” tasks. Topranking docked molecules can be further examined for drug potential. Quantum Mechanics Calculations NIH Pub. Chem Data. Base POVRay Parallel Rendering IU’s Varuna Data. Base Big Red (and the Tera. Grid) will also enable us to perform time consuming, multi-stepped Quantum Chemistry calculations on all of Pub. Med. Results go back to public databases that are freely accessible by the scientific community. CICC supports the NIH mission by combining state of the art chemical informatics techniques with • World class high performance computing • National-scale computing resources (Tera. Grid) • Internet-standard web services • International activities for service orchestration • Open distributed computing infrastructure for scientists world wide Indiana University Department of Chemistry, School of Informatics, and Pervasive Technology Laboratories 32
MLSCN Post-HTS Biology Decision Support Percent Inhibition or IC 50 data is retrieved from HTS Question: Was this screen successful? Workflows encoding plate & control well statistics, distribution analysis, etc Question: What should the active/inactive cutoffs be? Workflows encoding distribution analysis of screening results Question: What can we learn about the target protein or cell line from this screen? Workflows encoding statistical comparison of results to similar screens, docking of compounds into proteins to correlate binding, with activity, literature search of active compounds, etc Compounds submitted to Pub. Chem PROCESS CHEMINFORMATICS Grids can link data analysis ( e. g image processing developed in existing Grids), traditional Cheminformatics tools, as well as annotation tools (Semantic Web, del. icio. us) and enhance lead ID and SAR analysis A Grid of Grids linking collections of services at Pub. Chem ECCR centers MLSCN centers GRIDS 33
R Web Services 34
Why? £Need access to math and stat functionality £Did not want to recode algorithms £Wanted latest methods £Needed a distributed approach to computation ¤Keep computation on a powerful machine ¤Access it from a smaller machine 35
Why R? £Free, open-source £Many cutting edge methods avilable £Flexible programming language £Interfaces with many languages ¤Python ¤Perl ¤Java ¤C 36
The R Server £R can be run as a remote compute server ¤Requires the rserve package £Allows authenticated access over TCP/IP £Connections can maintain state £Client libraries for Java & C 37
R as a Web Service £On its own the R server is not a web service £We provide Java frontends to specific functionalities £The frontend classes are hosted in a Tomcat web container £Accessible via SOAP £Full Javadocs for all available WS’s 38
Flowchart 39
Functionality £Two classes of functionality ¤General functions ¥Allows you to supply data and build a predictive model ¥Sample from various distributions ¥Obtain scatter plots and hisotgram ¥Model development functions use a Java frontend to encapsulate model specific information 40
Functionality £Two classes of functionality ¤Model deployment ¥Allows you to build a model outside of the infrastructure ¥Place the final model in the infrastructure ¥Becomes available as a web service ¥Each model deployed requires its own front end class ¥In general, these classes are identical - could be autogenerated 41
Available Functionality £Predictive models - OLS, RF, CNN, LDA £Clustering - k-means £Statistical distributions £XY plot and scatter plots £Model deployment for single model types and ensemble model types 42
Deployed Models £Since deployed models are visible as web services we can build a simple web front end for them £Examples ¤NCI anti-cancer predictions ¤Ames mutagenicity predictions 43
Applications £ The R WS is not restricted to ‘atomic’ functionality £ Can write a whole R program ¤Load it on the R compute server ¤Provide a Java WS frontend £ Examples ¤Feature selection ¤Automated model generation ¤Pharmacokinetic parameter calculation 44
Data Input/Output £Most modeling applications require data matrices £Depending on client language we can use ¤SOAP array of arrays (2 D matrices) ¤SOAP array (1 D vector form of a 2 D matrix) ¤VOTables 45
Data Input/Output £Some R web services can take a URL to a VOTables document ¤Conversion to R or Java matrices is done by a local VOTables Java library £R also has basic support for VOTables directly ¤Ignores binary data streams 46
Interacting With R WS’s £Traditional WS’s do not maintain state £Predictive models are different ¤A model is built at one time ¤May be used for prediction at another time ¤Need to maintain state £State is maintained by serialization to R binary files on the compute server £Clients deal with model ID’s 47
Interacting with R WS’s £Protocol ¤Send data to model WS ¤Get back model ID ¤Get various information via model ID ¥Fitted values ¥Training statistics ¥New predictions 48
Cheminformatics at Indiana University School of Informatics David J. Wild djwild@indiana. edu Associate Director of Chemical Informatics & Assistant Professor Indiana University School of Informatics, Bloomington http: //djwild. info 49
Cheminformatics education at Indiana £ M. S. in Chemical Informatics ¤ 2 years, 36 semester hours ¤ Includes a 6 -hour capstone / research project ¤ Opportunity to work in Laboratory Informatics (IUPUI) or closely with Bioinformatics (IUB) ¤ Currently 9 students enrolled £ Ph. D. in Informatics, Cheminformatics Specialty ¤ 90 credit hours, including 30 hours dissertation research. Usually 4 years. ¤ Research rotations expose students to research in related areas ¤ Currently 4 students enrolled £ Graduate Certificate ¤ 4 courses, all available by Distance Education 50
Distance Education for Cheminformatics £ Uses Breeze + teleconference for live sharing of classes: all that is required is a P. C. and a telephone. Optional Polycom videoconferencing. £ Lectures are recorded for easy playback through a web browser £ Wiki or similar webpage for dissemination of course materials £ Also participate in CIC courseshare to give class at University of Michigan £ Of 75 students taking our courses since fall 51
Current research in the Wild lab £ Integration of cheminformatics tools and data sources ¤A web service infrastructure for cheminformatics ¤Compound information & aggregation web service and interface (“by the way box”) ¤An enhanced chatbot for exploting chemical information & web services ¤A semantically-aware workflow tools for cheminformatics ¤Data mining the NIH DTP tumor cell line database ¤Pub. Dock: a docking database for Pub. Chem 52
Current research in the Guha lab £ Predictive Modeling ¤Interpretation, validation, domain applicability ¤Generalization to other ‘models’ such as docking, pharmacophore etc ¤Integration of multiple data types ¤Addressing imbalanced and noisy datasets £ Analysis of Chemical Spaces ¤Quantify distributions in spaces ¤Investigation of density approaches ¤Applications to lead hopping, model domains £ Methods to summarize & compare data ¤Applications to HTS and smaller lead series type 53
Cheminformatics services Docking (FRED) 3 D structure generation (OMEGA) Filtering (FRED, etc) Database Services OSCAR 3 Postgre. SQL + g. Nova Fingerprints (BCI, CDK) Pub. Chem mirror Clustering (BCI) (augmented) Toxicity prediction Pub 3 D - 3 D structures (Tox. Tree) for Pub. Chem R-based predictive models Pub. Dock - Bound 3 D Similarity calculations structures (CDK) Compound-indexed Descriptor calculation journal article DB (CDK) Xiao Dong, Kevin E. Gilbert, Rajarshi Guha, Randy Heiland, Jungkee Kim, Marlon E. Pierce, Geoffrey C. Fox and NIH Human Tumor Cell David J. Wild, Web service infrastructure for chemoinformatics, Journal of Chemical Information and Modeling, 2007; 2 D 47(4) pp 1303 -1307 structure diagrams Line 54 (CDK) Local Pub. Chem mirror Cheminformatics web service infrastructure
RSC Project Prospect - what can we do with the information? £ www. projectprospect. org £ >100 papers marked up with SMILES/In. Ch. I (using OSCAR 3), plus Gene Ontology and Goldbook Ontology terms £ Created similarity searchable Postgre. SQL / g. Nova database with paper DOIs, SMILES, and ontology terms £ Web service and simple HTML interfaces for searching … “which papers reference compounds similar to this one in the scope of these ontological terms? ” 55
Greasemonkey / OSCAR script http: //cheminformatics. indiana. edu: 8080/Chem. GM/index. jsp 56
By the way… annotation (mock-up!) By the way… This compounds is very similar to a prescription drug, Tamoxifen. This compound is referenced in 20 journal articles published in the last 5 years Similar compounds are associated with the words “toxic” and “death” in 280 web pages It appears to be covered under 3 patents It has been shown to be active in 5 screens Computer models predict it to show some activity against 8 protein targets Here are some comments on this compound: David Wild: don’t take any notice of the computational models - they are rubbish 57
Cheminformatics aware simple lab notebook (mock up!) Plug-in allows structures to be drawn with the pen and cleaned up Some useful chemical reactions Iodoacetate a Iodoacetamide I-CH 4 COO- ICH 2 CONH 2 FIND INFO ABOUT THIS REACTION This may also react, chem favored by alkaline p. H Free text input can be converted to machine readable form by electrovaya …. Web service interface provides access to computation and searching. Page is marked up by what is possible Automatic detection of data fields (yield, etc) Where possible 58
Automatic workflow generation and natural language queries £ Develop service ontology using OWL-S or similar language 2 d similarity ¤Allows service interoperability, replacement and input/outut compatibility 3 D structures are compounds £ We can then use generic reasoning and 2 D -> 3 D network analysis tools to find paths from 2 D inputs to desired outputs structure crawler result £ Natural language can be parsed to inputs and P’phore dock search desired outputs £ Smart Clients <--> Agents <--> Services £ Possible “supercharged life science Google? ” 59 2 D structures 3 D search 3 D structures & complexes 2 D structures are compounds 3 D protein structure 3 D structures are compounds dock = bind
Скачать презентацию E-Chemistry and Web 2 0 Marlon Pierce mpierce cs
cf74ea41167dd3592c5426a407197221.ppt