HOME- R&D
- Bryn Mawr Conference
- Program & Schedule
- Location
- Screening Workshop
- Call for Contributions
- Bursary Awards
- Poster Session
- Sponsors & Exhibitors
- Program 2007
- ADMET
- Beger, R
- Cherkasov, A
- Contrera, J
- Helma, C
- Hopfinger, A
- Jamois, E
- Klon, A
- Madden, J
- Pelletier, D
- Poroikov, V
- Richard, A
- Tomaszewski, J
- Fragment-based Design
- SBDD
- Screening
- Structural Biology
- ADME/Tox Forum
- Program 2006
- Photo Gallery
- Hyderabad Conference
- Workshops & Training
- Program
- Exhibition
- Registration
- Jobs
- Contact
- Schedule
|
|
|
|
|
| About Joseph Contrera (FDA) |
|
Dr. Joseph F. Contrera is the Director of the Informatics and Computational Safety Analysis Staff at the FDA Office of Pharmaceutical Science in the Center for Drug Evaluation and Research (CDER). This group is responsible for developing and evaluating toxicology databases and computational toxicology predictive software for regulatory decision support and toxicological research. This effort is part of the FDA Critical Path Initiative. Information from toxicology databases developed by this group contributed to the development of International Conference on Harmonization (ICH) guidances for toxicology testing of human pharmaceuticals that are now applied internationally. Computational toxicology software modules developed by this group under collaborative research and development agreements (CRADA) with software developers are commercially available and are being used by the FDA, other regulatory agencies, the pharmaceutical industry and the NIH/NIDA Medications Discovery Program to support safety assessment and drug development.
Dr. Contrera received a B.A. degree in biology and chemistry from Washington Square College, New York University. He obtained an M.S. in neuropharmacology and physiology from NY University School of Medicine and a Ph.D. in physiology and endocrinology from NY University Graduate School. He received postdoctoral training in neuropharmacology at Yale University School of Medicine and John Hopkins University School of Medicine. After having served as Associate Professor of Physiology at the University of Maryland, he joined the FDA Division of Neuropharmacological Drug Products. Dr. Contrera was a supervisory pharmacologist in the FDA Division of Neuropharmacological Drug Products before joining the FDA/CDER Office of Pharmaceutical Science.
|
|
QSAR screening for carcinogenic potential using multiple models and software platforms
Joseph F. Contrera, PhD., Director, Informatics and Computational Safety Analysis, FDA Center for Drug Evaluation and Research, Office of Pharmaceutical Science, 10903 New Hampshire Avenue, Silver Spring, Maryland 20993, USA
The growing demand for computational methods to replace or reduce conventional animal testing has led to an increase in the number and variety of QSAR software applications. This, in turn leads to the evaluation and interpretation of the results of multiple predictive QSAR software programs. Different computational software programs employing different predictive algorithms can produce conflicting predictions of toxicological activity. Questions arise about how to best interpret such information. We have compared the results of two software used by regulatory agencies and the pharmaceutical industry, Multicase MC4PC and MDL-QSAR, using the same 1540 compound 2-year rodent carcinogenicity training data set in 10x10% leave-many-out internal validation studies. The MC4PC program uses a molecular fragment-based approach that reduces training data set chemical structures to all possible 2- to 10-atom fragments and then identifies those fragments primarily associated with active molecules (structural alerts). MDL-QSAR uses electrotopological structure descriptors, including atom-type and group-type E-State and hydrogen E-State indices, molecular connectivity indices, topological polarity, and counts of molecular features. Single software platform predictions and consensus predictions were generated, where information from more than one platform was combined. Although MC4PC and MDL-QSAR use different algorithms, their overall predictive performance was remarkably similar and the programs appeared complimentary, with compounds uncovered by one program covered by the other. Respectively, the sensitivity of MC4PC and MDL-QSAR was 61% and 63%, specificity was 71% and 75%, and concordance was 66% and 69%. Coverage for both programs was over 95%. Merging MC4PC and MDL-QSAR predictions improved the overall predictive performance. Consensus sensitivity increased to 67%, specificity to 84%, concordance to 76%.
Finally, user priorities may favor high specific rather than highly sensitive predictions, or vice versa. Consensus prediction rules can be adjusted to reflect the priorities of the user, so that greater emphasis may be placed on predictions from models with high sensitivity/low false negative rates or high specificity/low false positive rates.
|
|
|
|
|
|
|
|
|
|
|
