1. Building a Chemical Informatics Grid
Marlon Pierce
Community Grids Laboratory
Indiana University

2. Acknowledgments
CICC researchers and developers who contributed to this presentation:
Prof. Geoffrey Fox, Prof. David Wild, Prof. Mookie Baik, Prof. Gary Wiggins, Dr. Jungkee Kim, Dr. Rajarshi Guha, Sima Patel, Smitha Ajay, Xiao Dong
Thanks also to Prof. Peter Murray Rust and the WWMM group at Cambridge University
More info: and

3. Chemical Informatics and the Grid
An overview of the basic problem and solution

4. Chemical Informatics as a Grid Application
Chemical Informatics is the application of information technology to problems in chemistry.
Example problems: managing data in large scale drug discovery and molecular modeling
Building Blocks: Chemical Informatics Resources:
Chemical databases maintained by various groups
NIH PubChem, NIH DTP
Application codes (both commercial and open source)
Data mining, clustering
Quantum chemistry and molecular modeling
Visualization tools
Web resources: journal articles, etc.
A Chemical Informatics Grid will need to integrate these into a common, loosely coupled, distributed computing environment.

5. Problem: Connecting It Together
The problem is defining an architecture for tying all of these pieces into a distributed computing system.
A “Grid”
How can I combine application codes, web resources, and databases to solve a particular problem that interests me?
Specifically, how do I build a runtime environment that can connect the distributed services I need to solve an interesting problem?
For academic and government researchers, how can I do all of this in an open fashion?
Data and services can come from anywhere
That is, I must avoid proprietary infrastructure.

6. NIH Roadmap for Medical Research http://nihroadmap.nih.gov/
The NIH recognizes chemical and biological information management as critical to medical research.
Federally funded high throughput screening centers.
HTS assays per year on small molecules.
100,000’s of small molecules analyzed
Data published, publicly available through NIH PubChem online database.
What do you do with all of this data?

7. High-Throughput Screening
Testing perhaps millions of compounds in a corporate collection to see if any show activity against a certain disease protein

8. High-Throughput Screening
Traditionally, small numbers of compounds were tested for a particular project or therapeutic area
About 10 years ago, technology developed that enabled large numbers of compounds to be assayed quickly
High-throughput screening can now test 100,000 compounds a day for activity against a protein target
Maybe tens of thousands of these compounds will show some activity for the protein
The chemist needs to intelligently select the classes of compounds that show the most promise for being drugs to follow-up

9. Informatics Implications
Need to be able to store chemical structure and biological data for millions of data points
Computational representation of 2D structure
Need to be able to organize thousands of active compounds into meaningful groups
Group similar structures together and relate to activity
Need to learn as much information as possible (data mining)
Apply statistical methods to the structures and related information
Need to use molecular modeling to gain direct chemical insight into reactions.

10. The Solution, Part I: Web Services
Web Services provide the means for wrapping databases, applications, web scavengers, etc, with programming interfaces.
WSDL definitions define how to write clients to talk with databases, applications, etc.
Web Service messaging through SOAP
Discovery services such as UDDI, MDS, and so on.
Many toolkits available
Axis, .NET, gSOAP, SOAP::Lite, etc.
Web Services can be combined with each other into workflows
Workflow==use case scenario
More about this later.

11. Basic Architectures: Servlets/CGI and Web Services
Browser
Browser
GUI
Client
Web
Server
HTTP GET/POST
WSDL
SOAP
Web
Server
WSDL
Web
Server
WSDL
WSDL
SOAP
JDBC
JDBC DB DB

12. Solution Part II: Grid Resources
Many Grid tools provide powerful backend services
Globus: uniform, secure access to computing resources (like TeraGrid)
File management, resource allocation management, etc.
Condor: job scheduling on computer clusters and collections
SRB: data grid access
OGSA-DAI: uniform Grid interface to databases.
These have Web Service as well as other interfaces (or equivalently, protocols).

13. Solution, Part III: Domain Specific Tools and Standards -->More Services
For Chemical Informatics, we have a number of tools and standards.
Chemical string representations
SMILES, InChI
Chemistry Markup Language
XML language for describing, exchanging data.
JUMBO 5: a CML parser and library
Glue Tools and Applications
Chemistry Development Kit (CDK)
OpenBabel
These are the basis for building interoperable Chemical Informatics Web Services
Analogous situations exist for other domains
Astronomy, Geosciences, Biology/Bioinformatics

14. Solution Part IV: Workflows
Workflow engines allow you to connect services together into interesting composite applications.
This allows you to directly encode your scientific use case scenario as a graph of interacting services.
There are many workflow tools
We’ll briefly cover these later.
General guidance is to build web services first and then use workflow tools on top of these services.
Don’t get married to a particular workflow technology yet, unless someone pays you.

15. Solution Part V: User Interfaces
Web Services allow you to cleanly separate user interfaces from backend services.
Model-view-controller pattern for web applications
Client environments include
Grid and web service scripting environments
Desktop tools like Taverna and Kepler
Portlet-based Web portal systems
Typically, desktop tools like Taverna are used by power users to define interesting workflows.
Portals are for running canned workflows.

16. Next steps
Next we will review the online data base resources that are available to us.
Databases come in two varieties
Journal databases
Data databases
As we will discuss, it is useful to build services and workflows for automatically interacting with both types.

17. Online Chemical Journal and Data Resources

18. MEDLINE: Online Journal Database
MEDLINE (Medical Literature Analysis and Retrieval System Online) is an international literature database of life sciences and biomedical information.
It covers the fields of medicine, nursing, dentistry, veterinary medicine, and health care.
MEDLINE covers much of the literature in biology and biochemistry, and fields with no direct medical connection, such as molecular evolution.
It is accessed via PubMed.

19. PubMed: Journal Search Engine
PubMed is a free search engine offered by the United States National Library of Medicine as part of the Entrez information retrieval system.
The PubMed service allows searching the MEDLINE database.
MEDLINE covers over 4,800 journals published in the United States and more than 70 other countries primarily from 1966 to the present.
In addition to MEDLINE, PubMed also offers access to:
OLDMEDLINE for pre-1966 citations.
Citations to articles that are out-of-scope (e.g., general science and chemistry) from certain MEDLINE journals
In-process citations which provide a record for an article before it is indexed with MeSH and added to MEDLINE
Citations that precede the date that a journal was selected for MEDLINE indexing
Some life science journals

20. PubChem: Chemical Database
PubChem is a database of chemical molecules.
The system is maintained by the National Center for Biotechnology Information (NCBI) which belongs to the United States National Institutes of Health (NIH).
PubChem can be accessed for free through a web user interface.
And Web Services for programmatic access
PubChem contains mostly small molecules with a molecular mass below 500.
Anyone can contribute
The database is free to use, but it is not curated, so value of a specific compound information could be questionable.
NIH funded HTS results are (intended to be) available through pubchem.

21. NIH DTP Database Part of NIH’s Developmental Therapeutics Program.
Screens up to 3,000 compounds per year for potential anticancer activity.
Utilizes 59 different human tumor cell lines, representing leukemia, melanoma and cancers of the lung, colon, brain, ovary, breast, prostate, and kidney.
DTP screening results are part of PubChem and also available as a separate database.

22. Example screening results.
Positive results (red bar to right of vertical line) indicates greater than average toxicity of cell line to tested agent.

23. DTP and COMPARE
COMPARE is an algorithm for mining DTP result data to find and rank order compounds with similar DTP screening results.
Why COMPARE?
Discovered compounds may be less toxic to humans but just as effective against cancer cell lines.
May be much easier/safer to manufacture.
May be a guide to deeper understanding of experiments

24. Many Other Online Databases
Complementary protein information
Indiana University: Varuna project
Discussed in this presentation
University of Michigan: Binding MOAD
“Mother of All Databases”
Largest curated database of protein-ligand complexes
Subset of protein databank
Prof. Heather Carlson
University of Michigan: PDBBind
Provides a collection of experimentally measured binding affinity data (Kd, Ki, and IC50) exclusively for the protein-ligand complexes available in the Protein Data Bank (PDB)
Dr. Shaomeng Wang

25. The Point Is…
All of these databases can be accessed on line with human-usable interfaces.
But that’s not so important for our purposes
More importantly, many of them are beginning to define Web Service interfaces that let other programs interact with them.
Plenty of tools and libraries can simulate browsers, so you can also build your own service.
This allows us to remotely analyze databases with clustering and other applications without modifying the databases themselves.
Can be combined with text mining tools and web robots to find out who else is working in the area.

26. Encoding chemistry

27. Chemical Machine Languages
Interestingly, chemistry has defined three simple languages for encoding chemical information.
InChI, SMILES, CML
Can generate these by hand or automatically
InChIs and SMILES can represent molecules as a single string/character array.
Useful as keys for databases and for search queries in Google.
You can convert between SMILES and InChIs
OpenBabel, OELib, JOELib
CML is an XML format, and more verbose, but benefits from XML community tools

28. SMILES: Simplified Molecular Input Line Entry Specification
Language for describing the structure of chemical molecules using ASCII strings.

29. InChI: International Chemical Identifier
IUPAC and NIST Standard similar to SMILES
Encodes structural information about compounds
Based on open an standard and algorithms.

30. InChI in Public Chemistry Databases
US National Institute of Standards and Technology (NIST) - 150,000 structures
NIH/NCBI/PubChem project - >3.2 million structures
Thomson ISI - 2+ million structures
US National Cancer Institute(NCI) Database million structures
US Environmental Protection Agency(EPA)-DSSToX Database structures
Kyoto Encyclopaedia of Genes and Genomes (KEGG) database structures
University of California at San Francisco ZINC - >3.3 million structures
BRENDA enzyme information system (University of Cologne) - 36,000 structures
Chemical Entities of Biological Interest (ChEBI) database of the European Bioinformatics Institute structures
University of California Carcinogenic Potency Project structures
Compendium of Pesticide Common Names ( ) structures

31. Journals and Software Using InChI
Nature Chemical Biology.
Beilstein Journal of Organic Chemistry
Software
ACD/Labs ACD/ChemSketch.
ChemAxon Marvin.
SciTegic Pipeline Pilot.
CACTVS Chemoinformatics Toolkit by Xemistry, GmbH.

32. Chemistry Markup Language
CML is an XML markup language for encoding chemical information.
Developed by Peter Murray Rust, Henry Rzepa and others.
Actually dates from the SGML days before XML
More verbose than InChI and SMILES
But inherits XML schema, namespaces, parsers, XPATH, language binding tools like XML Beans, etc.
Not limited to structural information
Has OpenBabel support.

33. InChI Compared to SMILES
SMILES is proprietary and different algorithms can give different results.
Seven different unique SMILES for caffeine on Web sites:
[c]1([n+]([CH3])[c]([c]2([c]([n+]1[CH3])[n][cH][n+]2[CH3]))[O-])[O-]
CN1C(=O)N(C)C(=O)C(N(C)C=N2)=C12
Cn1cnc2n(C)c(=O)n(C)c(=O)c12
Cn1cnc2c1c(=O)n(C)c(=O)n2C
N1(C)C(=O)N(C)C2=C(C1=O)N(C)C=N2
O=C1C2=C(N=CN2C)N(C(=O)N1C)C
CN1C=NC2=C1C(=O)N(C)C(=O)N2C
On the other hand, some claim
SMILES are more intuitive for human readers.

34. A CML Example

35. Clustering Techniques, Computing Requirements, and Clustering Services
Computational techniques for organizing data

36. The Story So Far
We’ve discussed managing screening assay output as the key problem we face
Must sift through mountains of data in PubChem and DTP to find interesting compounds.
NIH funded High Throughput Screening will make this very important in the near future.
Need now a way to organize and analyze the data.

37. Clustering and Data Analysis
Clustering is a technique that can be applied to large data sets to find similarities
Popular technique in chemical informatics
Data sets are segmented into groups (clusters) in which members of the same cluster are similar to each other.
Clustering is distinct from classification,
There are no pre-determined characteristics used to define the membership of a cluster,
Although items in the same cluster are likely to have many characteristics in common.
Clustering can be applied to chemical structures, for example, in the screening of combinatorial or Markush compound libraries in the quest for new active pharmaceuticals.
We also note that these techniques are fairly primitive
More interesting clustering techniques exist but apparently are not well known by the chemical informatics community.

38. Non-Hierarchical Clustering
Clusters form around centroids.
The number of which can be specified by the user.
All clusters rank equally and there is no particular relationship between them.

39. Hierarchical Clustering
Clusters are arranged in hierarchies
Smaller clusters are contained within larger ones; the bottom of the hierarchy consists of individual objects in "singleton" clusters, while the top of it consists of one cluster containing all the objects in the dataset.
Such hierarchies can be built either from the bottom up (agglomerative) or the top downwards (divisive)

40. Fingerprinting and Dictionaries--What Is Your Parameter Space?
Clustering algorithms require a parameter space
Clusters defined along coordinate axes.
Coordinate axes defined by a dictionary of chemical structures.
Use binary on/off for fingerprinting a particular compound against a dictionary.

41. Cluster Analysis and Chemical Informatics
Used for organizing datasets into chemical series, to build predictive models, or to select representative compounds
Clustering Methods
Jarvis-Patrick and variants
O(N2), single partition
Ward’s method
Hierarchical, regarded as best, but at least O(N2)
K-means
< O(N2), requires set no of clusters, a little “messy”
Sphere-exclusion (Butina)
Fast, simple, similar to JP
Kohonen network
Clusters arranged in 2D grid, ideal for visualization

42. Limitations of Ward’s method for large datasets (>1m)
Best algorithms have O(N2) time requirement (RNN)
Requires random access to fingerprints
hence substantial memory requirements (O(N))
Problem of selection of best partition
can select desired number of clusters
Easily hit 4GB memory addressing limit on 32 bit machines
Approximately 2m compounds

43. Scaling up clustering methods
Parallelization
Clustering algorithms can be adapted for multiple processors
Some algorithms more appropriate than others for particular architectures
Ward’s has been parallelized for shared memory machines, but overhead considerable
New methods and algorithms
Divisive (“bisecting”) K-means method
Hierarchical Divisive
Approx. O(NlogN)

44. Divisive K-means Clustering
New hierarchical divisive method
Hierarchy built from top down, instead of bottom up
Divide complete dataset into two clusters
Continue dividing until all items are singletons
Each binary division done using K-means method
Originally proposed for document clustering
“Bisecting K-means”
Steinbach, Karypis and Kumar (Univ. Minnesota)
Found to be more effective than agglomerative methods
Forms more uniformly-sized clusters at given level

45. BCI Divkmeans Several options for detailed operation
Selection of next cluster for division
size, variance, diameter
affects selection of partitions from hierarchy, not shape of hierarchy
Options within each K-means division step
distance measure
choice of seeds
batch-mode or continuous update of centroids
termination criterion
Have developed parallel version for Linux clusters / grids in conjunction with BCI
For more information, see Barnard and Engels talks at:

46. Comparative execution times NCI subsets, 2
Comparative execution times NCI subsets, 2.2 GHz Intel Celeron processor
7h 27m
3h 06m
2h 25m
44m

47. Divisive K-means: Conclusions
Much faster than Ward’s, speed comparable to K-means, suitable for very large datasets (millions)
Time requirements approximately O(N log N)
Current implementation can cluster 1m compounds in under a week on a low-power desktop PC
Cluster 1m compounds in a few hours with a 4-node parallel Linux cluster
Better balance of cluster sizes than Wards or Kmeans
Visual inspection of clusters suggests better assembly of compound series than other methods
Better clustering of actives together than previously-studied methods
Memory requirements minimal
Experiments using AVIDD cluster and Teragrid forthcoming (50+ nodes)

48. Conclusions
Effective exploitation of large volumes and diverse sources of chemical information is a critical problem to solve, with a potential huge impact on the drug discovery process
Most information needs of chemists and drug discovery scientists are conceptually straightforward, but complex to implement
All of the technology is now in place to implement may of these information need “use-cases”: the four level model using service-oriented architectures together with smart clients look like a neat way of doing this
In conjunction with grid computing, rapid and effective organization and visualization of large chemical datasets is feasible in a web service environment
Some pieces are missing:
Chemical structure search of journals (wait for InChI)
Automated patent searching
Effective dataset organization
Effective interfaces, especially visualization of large numbers of 2D structures

49. Divisive K-Means as a Web Service
The previous exercise was intended to show that Divisive K-Means is a classic example of Grid application.
Needs to be parallelized
Should run on TeraGrid
How do you make this into a service?
We’ll go on a small tour before getting back to our problem.

50. Wrapping Science Applications as Services
Science Grid services typically must wrap legacy applications written in C or Fortran.
You must handle such problems as
Specifying several input and output files
These may need to be staged in
Launching executables and monitoring their progress.
Specifying environment variables
Often these have also shell scripts to do some miscellaneous tasks.
How do you convert this to WSDL?
Or (equivalently) how do you automatically generate the XML job description for WS-GRAM?

51. Generic Service Toolkit (GFAC) (G. Kandaswamy, IU and RENCI)
The Generic Service Toolkit can "wrap" any command-line application as an application service.
Given a set of input parameters, it runs the application, monitors the application and returns the results.
Requires no modification to program code.
Also has web user interface generating tools.
When a user accesses an application service, the user is presented with a graphical user interface (GUI) to that service.
The GUI contains a list of operations that the user is allowed to invoke on that service.
After choosing an operation, the user is presented with a GUI for that operation, which allows the user to specify all the input parameters to that operation.
The user can then invoke the operation on the service and get the output results.

52. OPAL (S. Krishan, SDSC)
Features include scheduling (using Globus and Condor/SGE) and security (using GSI-based certificates), and persistent state management.
The WSDL defines operations to do the following:
getAppMetadata: includes usage information, arbitrary application-specific metadata specified as an array of other elements,
e.g. description of the various options that are passed to the application binary.
launchJob: runs job with specified input and returns a Job ID.
queryStatus: returns status code, message, and URL of the working directory
getOutputs: returns the outputs from a job that is identified by a Job ID.
URLs for the standard output and error
Array of structures representing the output file names and URLs
getOutputAsBase64ByName: This operation returns the contents of an output file as Base64 binary.
destroy: This operation destroys a running job identified by a Job ID.
launchJobBlocking: This operation requires the list of arguments as a string, and an array of structures representing the input files.

53. Our Solution: Apache Ant Services
We’ve found using Apache Ant to be very useful for wrapping services.
Can call executables, set environment variables.
Lots of useful built-in shell-like tasks.
Extensible (write your own tasks).
Develop build scripts to run your application
You can easily call Ant from other Java programs.
So just write a wrapper service
We use both blocking (hold connection until return) and non-blocking version (suitable for long running codes).
In non-blocking case, “Context” web service is used for callbacks.

54. Flow Chart of SMILES to Cluster Partitioned of BCI Web Service
SMILES to DKM
SMILE
String
Makebits
Fingerprint
(*.scn)
DivKmeans
Cluster
Hierarchy
(*.dkm)
Generating
the best levels
Clustering
Fingerprints
Generating
Fingerprints
Dictionary
(Default)
New
SMILE
String
Extracting individual
cluster partitions
Extracted
Cluster
Hierarchy
(*.clu)
Optclus
RNNclus
One
Column
Process
Merge
Process
best
level

55. BCI Clustering Service Methods
Description
Input
Output
makebitsGenerate
Generate fingerprints from a SMILES structure
SMIstring
Fingerprint string
divkmGenerate
Cluster fingerprints with Divkmeans
SCNstring
Clustered Hierarchy
smile2dkm
Makebits + divkm
optclusGenerate
Generate the best levels in a hierarchy
DKMstring
Best partition cluster level
rnnclusGenerate
Extract individual cluster partitions
Indiv. cluster partitions
smile2ClusterPartitioned
Generate a new SMILES structure w/ extra col.
New SMILES structure

56. A Library of Chemical Informatics Web Services

57. All Services Great and Small
Like most Grids, a Chemical Informatics Grid will have the classic styles:
Data Grid Services: these provide access to data sources like PubChem, etc.
Execution Grid Services: used for running cluster analysis programs, molecular modeling codes, etc, on TeraGrid and similar places.
But we also need many additional services
Handling format conversions (InChI<->SMILES)
Shipping and manipulating tabular data
Determining toxicity of compounds
Generating batch 2D images
So one of our core activities is “build lots of services”

58. VOTables: Handling Tabular Data
Developed by the Virtual Observatory community for encoding astronomy data.
The VOTable format is an XML representation of the tabular data (data coming from BCI, NIH DTP databases, and so on).
VOTables-compatible tools have been built
We just inherit them.
SAVOT and JAVOT JAVA Parser APIs for VOTable allow us to easily build VOTable-based applications
Web Services
Spread sheet
Plotting applications.
VOPlot and TopCat are two

59. Document Structure of VOTable
<?xml version="1.0"?>
<VOTABLE version="1.1“ xmlns:xsi= xsi:noNamespaceSchemaLocation="
<RESOURCE >
<TABLE name="results">
<FIELD name=“CompoundName" ID="col1" datatype=“char" arraysize=“*”/>
<FIELD name=“ClustureNumber” ID="col2“ datatype=“int”/>
<DATA>
<TABLEDATA>
<TR><TD>Acemetacin</TD><TD>1</TD</TR>
<TR><TD>Candesartan</TD><TD>1</TD></TR>
<TR><TD>Acenocoumarol</TD><TD>2</TD></TR>
<TR><TD>Dicumarol</TD><TD>2</TD></TR>
<TR><TD>Phenprocoumon</TD><TD>2</TD></TR>
<TR><TD>Trioxsaken</TD><TD>2</TD></TR>
<TR><TD>warfarin</TD><TD>2</TD></TR>
</TABLEDATA>
</DATA>
</TABLE>
</RESOURCE>
</VOTABLE>
Compound
Name
Cluster
Number
Acemetacin 1
Candesartan
Acenocoumarol 2
Dicumarol
Phenprocoumon
Trioxsalen
Warfarin

60. mrtd1.txt – smiles representation of chemical compounds along with its properties

61. Taverna Client mrtd1.txt Tomcat Server WSDL VOTableGeneratorService
retrieveVOTableDocument
VOTableGeneratorService
votable.xml
VOPlot

62. Votable.xml : xml representation of mrtd1.txt file

63. VOPlot Application from generated votable
VOPlot Application from generated votable.xml file : Graph plotted on Mass (X–axis) and PSA (Y-axis)

64. Other Uses for VOTables
VOTables is a useful intermediate format for exchanging data between data bases.
Simple example: exchange data between VARUNA databases.
Each student in the Baik group maintains his/her on copy (sandbox purposes).
Often need to import/export individual data sets.
It is also good for storing intermediate results in workflows.
Value is not the format, but the fact that the XML can be manipulated programmatically.
Unions, subset, intersection operations

65. More Services: WWMM Services
Descriptions
Input
Output
InChIGoogle
Search an InChI structure through Google
inchiBasic
type
Search result in HTML format
InChIServer
Generate InChI
version
format
An InChI structure
OpenBabelServer
Transform a chemical format to another using Open Babel
inputData
outputData
options
Converted chemical structure string
CMLRSSServer
Generate CMLRSS feed from CML data
mol, title description link, source
Converted CMLRSS feed of CML data

66. CDK-Based Services Common Substructure
Calculates the common substructure between two molecules.
CDKsim
Takes two SMILES and evaluates the Tanimoto coefficient (ratio of intersection to union of their fingerprints).
CDKdesc
Calculates a variety of molecular and atomic descriptors for QSAR modeling
CDKws
Fingerprint generation
CDKsdg
Creates a jpeg of the compound’s 2D structure
CDKStruct3D
Generates 3D coordinates of a molecule from its SMILE

67. ToxTree Service
The Threshold of Toxicological Concern (TTC) establishes a level of exposure for all chemicals below which there would be no appreciable risk to human health.
ToxTree implements the Cramer Decision Tree approach to estimate TTC.
We have converted this into a service.
Uses SMILES as input.
Note the GUI must be separated from the library to be a service

68. toxTree

69. Taverna Workflow for Toxic Hazard Estimation

70. OSCAR3 Service
Oscar3 is a tool for shallow, chemistry-specific natural language parsing of chemical documents (i.e. 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.
There is a larger effort, SciBorg, in this area
This (like ToxTree) is potentially productively pleasingly parallelized.
It also has potentially very interesting Workflows

72. Use Cases and Workflows
Putting data and clustering together in a distributed environment.

73. Chemical Informatics as a Grid Problem
NIH-Funded experimental screening
NIH DTP and HTS projects are generating a wealth of raw data on small compounds.
Available in PubChem
Journal and chemical data sources often have public Web clients and GUIs.
But we need Web Service interfaces, not just Web interfaces.
These provide a programming interfaces for building both human and machine clients.
These need to be connected to computing resources for running clustering, data mining, and molecular modeling applications.
Excellent candidates for running on the TeraGrid
We can formulate scientific problems that map to inter-connections of Grid services.
This is generally called “Grid workflow” or “Service Orchestration”

74. ? SCIENTIST Computation
 All the compounds pass the Lipinksi Rule of Five and toxicity filters
Oracle Database (HTS)
 Compounds were tested against related assays and showed activity, including selectivity within target families
Excel Spreadsheet (Toxicity)
 One of the compounds was previously tested for toxicology and was found to have no liver toxicity
Oracle Database (Genomics)
? None of these compounds have been tested in a microarray assay
Word Document (Chemistry)
 Several of the compounds had been followed up in a previous project, and solubility problems prevented further development ?
Computation
 The information in the structures and known activity data is good enough to create a QSAR model with a confidence of 75%
SCIENTIST
Journal Article
 A recent journal article reported the effectiveness of some compounds in a related series against a target in the same family
“These compounds look promising from their HTS results. Should I commit some chemistry resources to following them up?”
External Database (Patent)
 Some structures with a similarity > 0.75 to these appear to be covered by a patent held by a competitor
Word Document (Marketing)
 A report by a team in Marketing casts doubt on whether the market for this target is big enough to make development cost-effective

75. Workflow, Services, and Science
Web Services work best as simple stateless services.
No implicit input, output, or interdependency of methods.
Services must be composed into interesting applications.
This is called workflow.
A good workflow ...
Is composed of independent services
Completely specifies an interesting science problem.

76. Some Open Source Grid Workflow Projects
UK e-Science Project’s Taverna
Scufl.xml scripting, GUI interface, works with Web Services.
Kepler
Works with Web services and the Globus Toolkit.
Condor DAGMan
Works over the top of Condor’s scheduler.
Extended by the GriPhyN Virtual Data System
Java CoGKit’s Karajan
XML workflow specification for scripting COG clients.
Works with GT 2 and 4.
Community Grids Lab’s HPSearch
JavaScript scripting, works with Web services.
Indiana Extreme Lab’s Workflow Composer
Jython, BPEL (soon) scripting

78. Finding compound-protein relationships
A 2D structure is supplied for input into the similarity search (in this case, the extracted bound ligand from the PDB IY4 complex)
A protein implicated in tumor growth is supplied to the docking program (in this case HSP90 taken from the PDB 1Y4 complex)
Correlation of docking results and “biological fingerprints” across the human tumor cell lines can help identify potential mechanisms of action of DTP compounds
The workflow employs our local NIH DTP database service to search 200,000 compounds tested in human tumor cellular assays for similar structures to the ligand. Client portlets are used to browse these structures
Once docking is complete, the user visualizes the high-scoring docked structures in a portlet using the JMOL applet.
Similar structures are filtered for drugability, and are automatically passed to the OpenEye FRED docking program for docking into the target protein.

79. HTS data organization & flagging
A tumor cell line is selected. The activity results for all the compounds in the DTP database in the given range are extracted from the PostgreSQL database
OpenEye FILTER is used to calculate biological and chemical properties of the compounds that are related to their potential effectiveness as drugs
The compounds are clustered on chemical structure similarity, to group similar compounds together
The compounds along with property and cluster information are converted to VOTABLES format and displayed in VOPLOT

80. Use Case: Which of these hits should I follow up?
An HTS experiment has produced 10,000 possible hits out of a screening set of 2m compounds. A chemist on the project wants to know what the most promising series of compounds for follow-up are, based on:
Series selection  cluster analysis
Structure-activity relationships  modal fingerprints/stigmata
Chemical and pharmacokinetic properties mitools, chemaxon
Compound history gNova / PostgreSQL
Patentability  BCI Markush handling software
Toxicity
Synthetic feasibility
+ requires visualization tools!

81. A Workflow Scenario: HTS Data Organization and Flagging
This workflow demonstrates how screening data can be flagged and organized for human analysis.
The compounds and data values for a particular screen are retrieved from the NIH DTP database and then are filtered to remove compounds with reactive groups, etc.
A tumor cell line is selected. The activity results for all the compounds in the DTP database in the given range are extracted from the PostgreSQL database
OpenEye FILTER is used to calculate biological and chemical properties of the compounds that are related to their potential effectiveness as drugs
ToxTree is used to flag the potential toxicities of compounds.
Divkmeans is used to add a column of cluster numbers.
Finally, the results are visualized using VOPlot and the 2D viewer applet.

82. Web Services

83. Example plots of our workflow output using VOPlot and VOTables

84. SMILES + ID + + Cluster # + Data
SMILES + ID + Data
Fingerprint
Generator
BCI Makebits
Cluster
Analysis
BCI Divkmeans
NIH Database
Service
PostgreSQL
CHORD
SMILES + ID
Cluster
Membership
Table
Management
VoTables 
Fingerprints 
Cluster the compounds in the NIH DTP database by chemical structure, then choose representative compounds from the clusters and dock them into PDB protein files of interest
SMILES + ID + + Cluster # + Data
Plot
Visualizer
VoPlot
Docking
Selector
Script
3D Visualizer
JMOL  
SMILES + ID
Docking
OpenEye FRED
2D-3D
OpenEye OMEGA
PDB Database
Service
Docked Complex 
MOL File 
PDB Structure + Box

85. Use Case: Are there any good ligands for my target?
A chemist is working on a project involving a particular protein target, and wants to know:
Any newly published compounds which might fit the protein receptor site  gNova / PostgreSQL, PubChem search, FRED Docking
Any published 3D structures of the protein or of protein-ligand complexes  PDB search
Any interactions of compounds with other proteins  gNova / PostgreSQL, PubChem search
Any information published on the protein target  Journal text search

86. Use Case: Who else is working on these structures?
A chemist is working on a chemical series for a particular project and wants to know:
If anyone publishes anything using the same or related compounds ~ PubChem search
Any new compounds added to the corporate collection which are similar or related  gNova CHORD / PostgreSQL
If any patents are submitted that might overlap the compounds he is working on ~ BCI Markush handling software
Any pharmacological or toxicological results for those or related compounds  gNova CHORD / PostgreSQL, MiToolkit
The results for any other projects for which those compounds were screened  gNova CHORD / PostgreSQL, PubChem search

87. VARUNA – Towards a Grid-based Molecular Modeling Environment
A brief overview of Prof. Mookie Baik’s VARUNA project.

88. Chemical Informatics in Academic Research?
Industrial Research: Target Oriented
Not bound to a specific molecular system
Not bound to a method
Not concerned with generality
Aware of Efficiency
Aware of Overall Cost
Aware of Toxicity
Concerned about Formulations
Cares about active MOLECULES
Academic Research: Concept Oriented
Specialized on few molecular families
Method Development is important
Obsessed with generality
Does not care much about efficiency
Cost is unimportant
Often can’t even assess for Toxicity
Formulation is a minor issue
Cares mostly about REACTIONS, i.e. ways to GET to a molecule

89. AutoGeFF, Varuna and Workflows
Metalloproteins are extremely important in biochemical processes
Understanding their chemistry is difficult
To add value to the small molecule DB’s (PubChem, etc.), we must somehow connect them to PDB’s, BindMOAD, etc.
By extending Varuna’s functionality to handling, storing Metalloproteins, we could provide a connection

90. Automatic Generator of ForceFields (AutoGeFF)
Developing a service that can take ANY
drug-like molecule (from PubChem, for example)
metal complexes
metalloenzymes (from PDB, for example)
unnatural or functionalized amino acids, nucleobases (from in-house db)
for which molecular mechanics force fields are not available and automatically generate FF’s based on
High level Quantum Simulations (using Varuna as a Web service)
for Sophisticated Molecular Mechanics Simulations
First Step: Coding of a specialized Prototype that can reproduce our manually derived novel force fields for Cu-Ab Alzheimer’s Disease as a Proof-Of-Principles Study.

91. Automatic Quantum Mechanical Curation of Structure Data
Chemical Research logic is often driven by molecular structure
Large-scale, small molecule DB’s (such as PubChem) have low-resolution structure data
Often key properties are not consistently available:
e.g.: Rotation-barriers, Redox Potentials, Polarizabilities, IR frequencies, reactivity towards nucleophiles
QM web-services will provide tools for generating high-resolution data
that will curate the results of traditional ChemInfo studies
allow for combinatorial computational chemistry
access a database of modeling data

92. Prototype-Project: Controlling the TGFb pathway
Simulations
in-house Molecules in Varuna
AutoGeFF
VARUNA
Conceptual
Understanding of TGFb
Inhibition
Inactive TGFb
Active TGFb
With inhibitor
1IAS
Questions:
- What molecular feature controls inhibitor binding?
- How do mutations impact binding?
PubChem
Experiments in the Zhang
Lab
PDB

93. Consequences for ChemInfo Design for Academia
TWO Strategies are needed:
Making traditional ChemInfo tools that are often available in commercial research available to Academia is in principle straightforward.
New ChemInfo Tools that are CONCEPT centered and include REACTIONS in addition to MOLECULES must be developed.
Our approach: Development of
(a) Quantum Chemical Database
(b) Molecular Modeling Database
Harness the power of recent advances in Molecular Modeling (QM, QM/MM, MM, MD) through information management.
Data-depository for Quantum Chemical Data including both Properties & Mechanisms

94. QM Calculation Workflow

95. More Information Contact me: mpierce@cs.indiana.edu
Most of this was taken from our CICC project. See
Note we’ve found wikis to be extremely useful and fun to use for maintaining collaborative web sites.
See also and for other examples using Media Wiki.
Many elements of our approach are based on Prof. Peter Murray Rust’s group’s approach.
WWMM Wiki: wwmm.ch.cam.ac.uk/wikis/wwmm/index.php
SourceForge Project Site

96. Additional Slides

97. Use Case - CICC Which of these hits should I follow up?
An MLI HTS experiment has produced 10,000 possible hits out of a screening set of 2m compounds. A chemist at another laboratory wants to know if there are any interesting active series she might want to pursue, based on:
Structure-activity relationships
Chemical and pharmacokinetic properties
Compound history
Patentability
Toxicity
Synthetic feasibility

98. CICC Web Services I BCI Clustering ToxTree
Provides Bernard Chemical Information (BCI) clustering packages
A module of the workflow for HTS data organization and flagging
Status:
Added URL output support to the previous solid prototype (Multi-user durable)
Taverna Beanshell Scripting for data format adjusting (e.g. Filtering out the head part listing column names)
To do: Evaluating the URI(URL) based workflow design
ToxTree
Estimates toxic hazard by applying a decision tree approach
Status: A test prototype producing the level of toxicity in a brief or verbose explanation against a SMILE structure
To do:
Refining the Web service for cluster input and external property support
The Taverna Beanshell scripting for data merging not used in some modules

99. CICC Web Services II Workflow for HTS data organization and flagging
Demonstrates how screening data can be flagged and organized for human analysis
Status: Individual modules except the visualization are in prototype
To do:
Defining at least XML schema or DTD for the workflow data (at most the Ontology)
Redefining current workflow model to reflect the new feature of Taverna 1.4 supporting complex data structures and the provenance plugin
Other Planed Web Services
Open Source Chemistry Analysis Routines (OSCAR)
Extracts chemical information from text and produces an XML instance highlighting the chemical information
A module of the PMR workflow
Status: OSCAR3 is available and works fine as a Java application
To do: Studying XML instances for extracting chemical names
InfoChem’s SPRESI Web Service
Provides access to the SPRESI molecule database
Status: Perl scripts for accessing SPRESI Web Service
To do: Developing a Web service wrapper to utilize InfoChem’s SPRESI Web Service

100. BCI Clustering URL Service Methods
Description
Input
URLOutput
makebitsURLGenerate
Generate fingerprints from a SMILES structure
SMIstring
Fingerprint and program output
divkmURLGenerate
Cluster fingerprints with Divkmeans
SCNstring
DKM data and program output
smile2dkmURL
Makebits + divkm
All SMI, DKM and std. outputs
optclusURLGenerate
Generate the best levels in a hierarchy
DKMstring
Best data and program output
rnnclusURLGenerate
Extract individual cluster partitions
New partition and std. output
smile2ClusterPartitionedURL
Generate a new SMILES structure w/ extra col.
All intermediate data and output

101. Workflow for smile2ClusterPartitionedURL

102. Workflow for Toxic Hazard in Verbose

103. Diagram of Workflow2
Input/Output
Web Services
Beanshell Scripting

104. Informatics
Informatics is the discipline of science which investigates the structure and properties (not specific content) of scientific information, as well as the regularities of scientific information activity, its theory, history, methodology and organization. The purpose of informatics consists in developing optimal methods and means of presentation (recording), collection, analytical-synthetic processing, storage, retrieval and dissemination of scientific information.
A. I. Mikhailov, A. I. Chernyi, R. S. Gilyarevskii (1967) “Informatics -- New Name of the Theory of Scientific Information”

105. Chemical informatics is …
More usually know as chemoinformatics or cheminformatics
Very differently defined, reflecting its cross-disciplinary nature
Librarian
Chemist (synthetic, medicinal, theoretical)
Biologist / Bioinformatician
Molecular modeler
Pharmaceutical or Chemical Engineer
Computer Scientist / Informatician

106. More definitions
Computational Chemistry – The application of mathematical and computational methods to particularly to theoretical chemistry
Molecular Modeling – Using 3D graphics and optimization techniques to help understand the nature and action of compounds and proteins
Computer-Aided Drug Design – The discipline of using computational techniques (including chemical informatics) to assist in the discovery and design of drugs.

107. Traditional areas of application
Pharmaceutical & life science industry
particularly in early stage drug design
Databases of available chemicals
Electronic publishing
including searchable chemical structure information in journals, etc.
Government and patent databases

108. The –ics so far (1960’s to present) …
How do you represent 2D and 3D chemical structures?
Not just a pretty picture
How do you search databases of chemical structures?
Google doesn’t help (much, but it might do soon…)
How do you organize large amounts of chemical information?
How do you visualize chemical structures & proteins?
Can computers predict how chemicals are going to behave
… in the test tube?
… in the body?

109. Current trends & hot topics
The decorporatization of chemical informatics (PubChem, MLI, eScience, open source)
Service-oriented architectures
Packaging & processing large volumes of complex information for human consumption
Integration with other –ics (bioinformatics, genomics, proteomics, systems biology)

110. Main players (Commercial)
MDL
Tripos, inc.
Accelrys
Daylight CIS, inc.

111. Main players (Academia)
“Pure” Chemoinformatics
University of Sheffield, UK (Willett / Gillet)
Erlangen, Germany (Gasteiger)
Cambridge Unilever Center
Indiana University School of Informatics
Related (computational chemistry, etc.)
UCSF (Kuntz)
University of Texas (Pearlman)
Yale (Jorgensen)
University of Michigan (Crippen)

112. “Traditional” Journals
Journal of Chemical Information & Modeling (formerly JCICS)
Journal of Computer-Aided Molecular Design
Journal of Molecular Graphics and Modeling
Journal of Computational Chemistry
Journal of Chemical Theory and Computation
Journal of Medicinal Chemistry

113. “Informal” publications
Network Science (online)
Chemical & Engineering News
Drug Discovery Today
Scientific Computing World
Bio-IT World

114. CINF-L Distribution List
Chemical Information Sources Discussion List
Created by Gary Wiggins at IUB

115. Yahoo! Chemoinformatics Discussion List
Job postings
Ideas exchange
Questions
Industry – Student connections
All students encouraged to join
Open to others
To join, go to
Or send an to

116. Open Source / Free Software
Blue Obelisk -
InChI -
JMOL –
FROWNS -
OpenBabel -
CML -
CDK -
MMTK -

117. Example 2 3D Visualization & Docking
3D Visualization of interactions between compounds and proteins
“Docking” compounds into proteins computationally

118. 3D Visualization
X-ray crystallography and NMR Spectroscopy can reveal 3D structure of protein and bound compounds
Visualization of these “complexes” of proteins and potential drugs can help scientists understand the mechanism of action of the drug and to improve the design of a drug
Visualization uses computational “ball and stick” model of atoms and bonds, as well as surfaces
Stereoscopic visualization available

119. Docking algorithms
Require 3D atomic structure for protein, and 3D structure for compound (“ligand”)
May require initial rough positioning for the ligand
Will use an optimization method to try and find the best rotation and translation of the ligand in the protein, for optimal binding affinity

120. Genetic Algorithms
Create a “population” of possible solutions, encoded as “chromosomes”
Use “fitness function” to score solutions
Good solutions are combined together (“crossover”) and altered (“mutation”) to provide new solutions
The process repeats until the population “converges” on a solution

121. Traditional Workflow of Molecular Modeling
FORTRAN Code,
Scripts, Visualization Code
Supercomputer
Researcher
Hard Drive
Directory Jungle
Chemical
Concepts
Highly inefficient workflow (no automation)
Knowledge is human bound (grad student leaves and projects dies)
Incorporation with other DB’s is done in Researcher’s head
Experiments

122. Varuna – a new environment for molecular modeling
Chemical
Concepts
Researcher
Experiments
Chem-Grid
Simulation Service FORTRAN Code,
Scripts
DB Service Queries, Clustering, Curation, etc.
Reaction DB QM
Database
PubChem, PDB, NCI, etc.
QM/MM
Database
Supercomputer

123. Tools for mining the data
Tripos Benchware HTS Dataminer (formerly SAR Navigator),