Skip to main content

Clinical Genomic Analysis Workshop 2012

Sunday May 13, 2012
Organized by IBM Research – Haifa and Edmond J. Safra Center for Bioinformatics at Tel-Aviv University

You are cordially invited to participate in a one-day leadership seminar on clinical genomic analysis, to be held Sunday, May 13, 2012, from 9:30 to 17:00 at the IBM research lab, on the University of Haifa campus in Haifa, Israel. Lunch and light refreshments will be served. Participation is free.

This full-day workshop will provide a forum for the research and development communities from both academia and industry to share their work, exchange ideas, and discuss issues, problems, and works-in-progress. The forum will also address future research directions and trends in the area of personalized healthcare and the use of "omics" techonology for optimizing the individual care.

This year, we will devote a panel discussion to a current trend in the pharma world—real-world evidence (RWE).

Student authors are asked to submit an abstract for poster presentation before May 10, 2012.

Please confirm your participation by May 10, via the seminar registration page.


Table header results




Opening Remarks,
Rick Kaplan, IBM Israel Country General Manager


Personalized Medicine,
Gabi Barbash, Director General, Tel Aviv Sourasky Medical Center (Ichilov)

Dr. Gabriel Barbash has been the Director General of the Tel Aviv Sourasky Medical Center since 1993. He served as the Director General (Surgeon General) of the Ministry of Health from 1996 to 1999. Since 1995, Dr Barbash has been the Chairman of the national project of developing and implementing a SAP management and clinical information system for the 11 governmental medical centers comprised of more than 14,000 users.
From 1998 to 2001, Dr. Barbash served as the Chairman of the Israeli National Transplant Center and reorganized the system of organ harvesting in Israel, doubling the number of organ transplantations nationwide.
Dr. Barbash was Israel's national coordinator and principal investigator for numerous multi-center, international cardiology studies in which each department of cardiology in Israel took part. He has published more than 80 original papers, mainly in the fields of diagnosis, risk assessment, and treatment of acute myocardial infarction. In 2001, Dr. Barbash was appointed Professor of Epidemiology and Preventive Medicine in the Sackler School of Medicine, Tel Aviv University.
Dr. Barbash is a graduate of the Hadassah Medical School of the Hebrew University, Jerusalem, and is board certified in Internal Medicine, Medical Management and Occupational Medicine. He also holds a master's degree in Public Health (MPH), specializing in Health Policy and Management, from the School of Public Health at Harvard University. Dr. Barbash is a visiting professor in the Mailman School of Public health at Columbia University, New York, where with the US Ministry of Health Agency for Health Research Quality (AHRQ) he researches the diffusion of medical technologies.


Detecting Disease-Associated Genes with Rare Variants using Pooled, Low-Coverage Sequencing,
Oron Navon, Tel Aviv University

The development of modern DNA sequencing technologies now allows researchers to study both common and rare genetic variants involved in disease. However, power to detect the effects of rare variants, within the constraints of realistic budgets, is very low. To increase power, several methods have been developed to group together variants by gene or genomic region, and test for association between a disease and the set of variants within a region. Still, detecting subtle associations currently requires studies including hundreds or thousands of individuals, which is prohibitively costly utilizing current sequencing technologies. Two promising cost-reducing strategies are low coverage sequencing, which produces more error- prone data at significantly lower cost, and DNA pooling, where a pool containing a mixture of DNA samples from multiple individuals is sequenced in one run of the sequencing platform. Current methods, however, cannot be applied directly to such data, as they require individual genotypes, which are lost in pooling, and are prone to errors in low coverage sequencing.
This work describes two novel methods for the analysis of rare Single-Nucleotide Variations (SNVs) in sequencing data from DNA pools, characterized by low coverage and sequencing error. The novel methods are shown by computer simulation to outperform previous methods, even in the case of high coverage and with- out pooling. Through analysis of real pooled sequencing data from a study of non-Hodgkin's lymphoma, the sequencing error rate and the accuracy of pooling are estimated by comparing sequencing data to previously obtained whole-genome genotyping data on the same samples. Lastly, by comparing different study designs based on the parameters from the real data, it is shown that for a given budget, there exists an ideal pool size which dictates the number of cases to collect in order to maximize power to detect associations.

Oron Navon is fascinated with the utilization of big data for solving real-world problems, especially in biology and bioinformatics. He received his B.Sc. in Computer Science in the Bioinformatics Track at Ben-Gurion University in Beer-Sheva, and has recently completed the requirements toward an M.Sc. degree in Life Sciences in the Edmond J. Safra Bioinformatics Program at Tel-Aviv University. He currently works at AdiMap, Ltd., a start-up company which analyzes large amounts of consumer and product data to match online users with products they need.




Regulation of mammalian life-span by SIRT6,
Haim Cohen, Bar-Ilan University

For more than 70 years, it has been known that dietary restricted (DR) diet slows the rate of aging and extends the lifespan of many organisms. Moreover, rodents fed a DR diet exhibit a spectrum of phenotypes that are the direct opposite of the metabolic syndrome, including improved glucose tolerance; decreased total body fat, LDL cholesterol, free fatty acids (FFA) and triglycerides; and increased HDL cholesterol. Recently, we showed that the protein levels of SIRT6, one of the seven mammalian sirtuin deacetylases SIRT1 to 7, increase in rodents fed with DR. The highly conserved sirtuin deacetylases¬ were shown to regulate lifespan in lower organisms and to regulate glucose and fat homeostasis and age-related metabolic diseases in mammals. The findings that SIRT6 levels increase upon DR, suggest that SIRT6 is involved in the beneficial effect of DR and that overexpression of SIRT6 might mimic DR. To explore the role of SIRT6 in metabolic stress, wild type and mice overexpressing exogenous SIRT6 (MOSES) were fed a high fat diet. In comparison to their wild-type littermates, MOSES mice accumulated significantly less visceral fat, LDL-cholesterol, and triglycerides. MOSES mice displayed enhanced glucose tolerance along with increased glucose-stimulated insulin secretion. Given that these metabolic defects are known to be associated with aging we followed the lifespan of MOSES mice. Here we show that male, but not female, transgenic mice overexpressing Sirt6 have a significantly longer lifespan than wild-type mice. Gene expression analysis revealed significant differences between male MOSES mice and male wild-type mice: transgenic males displayed lower serum levels of insulin-like growth factor 1 (IGF1), higher levels of IGF-binding protein 1 and altered phosphorylation levels of major components of IGF1 signalling, a key pathway in the regulation of lifespan. These results demonstrate a protective role for SIRT6 against the metabolic consequences of diet-induced obesity and show the regulation of mammalian lifespan by a sirtuin family member.

Dr. Haim Cohen is is a senior lecturer at the Mina & Everard Goodman Faculty of Life Sciences, Bar-Ilan University, Ramat-Gan, Israel.


Finding Most Likely Haplotypes in General Pedigrees through Parallel Branch and Bound Search,
Rina Dechter, University of California Irvine

The maximum likelihood haplotype problem consists of finding a joint haplotype configuration for all members of a given pedigree which maximizes the probability of data (e.g., phenotypes of individuals and partial unordered genotype information at some marker loci). It can be shown that general pedigrees can be encoded as Bayesian networks, where the common Most Probable Explanation (MPE) query corresponds to finding the most likely haplotype configuration (Fishelson & Geiger 2002; Fishelson, Dovgolevsky, & Geiger 2005). In this talk I will present a strategy for grid parallelization of a state of the art Branch and Bound algorithm for MPE where independent worker nodes solve subproblems concurrently. The crucial issue of load balancing is addressed by estimating subproblem complexity through learning a regression model, using a variety of subproblem features (structural as well as dynamic, cost function-based). Experimental evaluation of the parallel scheme on several hundred CPUs yields promising results, solving a number of very hard pedigree problem instances with good parallel speedup, compared against the leading sequential Branch and Bound algorithm. Our scheme is currently implemented and being used within SUPERLINK ONLINE SNP developed by Dan Geiger's group in the Technion.
More information can be found at:

Rina Dechter is a professor of Computer Science at the University of California, Irvine. She received her PhD in Computer Science at UCLA in 1985, an MS degree in Applied Mathematic from the Weizmann Institute and a B.S in Mathematics and Statistics from the Hebrew University, Jerusalem. Her research centers on computational aspects of automated reasoning and knowledge representation including search, constraint processing and probabilistic reasoning.

Professor Dechter is an author of Constraint Processing published by Morgan Kaufmann, 2003, has authored over 100 research papers, and has served on the editorial boards of: Artificial Intelligence, the Constraint Journal, Journal of Artificial Intelligence Research and Logical Method in Computer Science (LMCS). She was awarded the Presidential Young investigator award in 1991, is a fellow of the American association of Artificial Intelligence since 1994, was a Radcliffe Fellowship 2005-2006 and received the 2007 Association of Constraint Programming (ACP) research excellence award. She is currently a co-editor in chief of the Artificial Intelligence Journal.


Panel: Real world evidence—a current trend in the pharma world,
Nava Siegelmann-Danieli, Lior Soussan-Gutman, Eran Dolev, Moderator: Michal Rosen-Zvi

Nava Siegelmann-Danieli, M.D.
Director of Oncology Service line at the Maccabi Health Services
Board certified in Internal Medicine and Oncology
Graduated the Hebrew University and Hadassah Medical School
Fellowship in Medical Oncology at the Fox Chase Cancer Center in PA USA
Served as a senior oncology physician at Fox Chase Cancer Center and at the Gesisinger Medical Center in PA USA, and at the Rambam and Assaf Harofeh hospitals  in Israel.
Recipient of the ASCO (American society of clinical oncology) Young investigator award as well as research grants from the AACR (American association of cancer research) and NIH.

Dr. Soussan-Gutman Managing Director of Oncotest-TEVA business unit in Teva Pharmaceuticals Inc. The unit provides an extended basket of services in molecular oncology, designed to tailor treatment to the unique characteristics of both patients and their disease. Dr. Soussan-Gutman was the founder and CEO of Oncotest Ltd., a company offering specialized cancer diagnostic service to the oncology community based on a network of laboratories around the world.
Dr. Soussan-Gutman founded Oncotest Ltd. in 1998 and in 2003 Oncotest's activities were purchased by Teva Pharmaceutical Industries in Israel and became the Oncotest-Teva service unit, managed by Dr. Lior Soussan-Gutman.
Dr. Soussan-Gutman holds a Ph.D. in Neurobiochemistry from the Tel-Aviv University and she did Post-doctorate research in molecular biology at the Weizmann Institute of Science.

Professor Eran Dolev is Professor of Medical Sciences & History of Medicine, Sackler School of Medicine, Tel-Aviv University. He obtained MD in 1965 from the Hebrew University School of Medicine, Jerusalem.
In 1979-1983  Prof Dolev served as Surgeon General,  Israel Defence Forces.
In 1990-2005  Prof Dolev served as the head of the Department of Internal Medicine,  Tel-Aviv Medical Center.
In 1996-2001  he served as chairman of the Israel Medical Association Ethical Committee.His main fields of research include Mineral Metabolism & Bone Diseases, Bio-Medical Ethics and Military Medicine & History of Military Medicine.

Dr. Michal Rosen-Zvi is a research staff member at IBM Research - Haifa in Israel. She holds a PhD in physics and completed her postdoctoral studies at UC Berkeley, UC Irvine and the Hebrew University. During that period, Michal worked with colleagues on developing novel machine learning methods. Towards the end of 2005 she joined IBM Research - Haifa, where she is now manager of the machine learning and data mining group. Michal has published more than 30 papers in leading journals and conferences, including ISMB conference, HIV medicine journal, and more. She serves as a Program Committee member and reviewer at top machine learning and bioinformatics conferences, including ICML, AISTAT, UAI, NIPS, KDD and ISMB/ECCB, and as a reviewer for journals such as Journal of Machine Learning Research, Machine Learning Journal, Bayesian Analysis Journal, IEEE Transactions on Computers, Journal of Artificial Intelligence Research (JAIR), bioinformatics, and more. Dr. Rosen-Zvi also gave talks in many different forums on optimization, machine learning and bioinformatics, and a course at Tel-Aviv University on applying Bayesian network methods to the clinical domain.




Shared neuronal pathways affected by common and rare variants in autism spectrum disorders,
Sagiv Shifman, Hebrew University

Recent studies into the genetics of Autism spectrum disorders (ASD) have implicated both common and rare variants, including de-novo mutations, as risk factors for ASD. However, how much of the genetic risk can be attributed to rare versus common alleles is unknown. Furthermore, the genes already known to be disrupted by rare variants still account for only a small proportion of the cases due to their rarity in the affected population. This genetic heterogeneity constitutes a considerable obstacle to establishing a thorough understanding of the etiology of ASD. To shed new light into the respective involvement of common and rare variation in autism, we constructed a gene co-expression network based on a widespread survey of gene expression in the human brain. The constructed network included modules associated with specific cell types and processes. These include two neuronal modules that were found to be enriched for both rare and common variations that are potentially associated with ASD risk. The enrichment for common variations in these modules was validated in two independent cohorts. The modules showing the highest enrichment for rare and common variants in ASD included highly connected genes that are involved in synaptic and neuronal plasticity and that are expressed in areas associated with learning and memory and sensory perception. Additionally, we found that the level of expression of the most connected genes in this module increases in the brain during fetal development, with a peak during the first year of life. Taken together, these results suggest a common role for rare and common variations in autism, and illustrate how rare and de novo mutations, in conjunction with common variations, can act together to perturb key pathways involved in neuronal processes, and specifically neuronal plasticity. Furthermore, the modules found in this study may serve as starting points for designing potential therapeutic interventions for ASD.


Avoiding the Obvious: A Clustering Method for Revealing Multiple Meaningful Partitions in Aggregated Medical Data,

Ruty Rinott, IBM Research - Haifa

Clustering allows revealing the hidden structure of the examined data. However, often, the most dominant clusters in medical data are of little interest, pertaining, for example, to the patient cohorts in which the data was collected, while masking more intriguing signals of potential clinical importance. Here we present a clustering method that relies on a simple information theoretic principle, and is capable of detecting multiple meaningful partitions of a single dataset. Our method works by iteratively finding additional clustering partitions of the entire data, that group together similar data instances while directly controlling the level of dependency of the obtained new partition with previously extracted partitions. We prove the merits of our method by testing it on two clinical datasets and show that while a standard clustering procedure groups the data by obvious yet irrelevant features; our method succeeds in extracting additional partitions that are clinically meaningful.
Joined work with Lavi Shpigelman, Oliver Keller, and Noam Slonim.

Ruty Rinott is a Research Staff Member in the Machine Learning and Data Mining group, at the Analytics department at Haifa research labs. She received her B.Sc. and M.Sc. degrees in computer science and computational biology from the Hebrew University of Jerusalem in 2008 and 2010, respectively. She joined IBM research in 2010, and since then has worked mainly on machine learning related project in the domains of medical informatics and bio-informatics.


Break and Poster session


Quantifying the "Clinical" Predictive Capacity of Genomes,
Saharon Rosset, Tel Aviv University

In a recent paper, Vogelstein and colleagues attempted to quantify the range of practical clinical utility of whole-genome information. Their results led the NY Times to declare "Study Says DNA Power to Predict Illness Is Limited". We survey their methodology and results, and describe our detailed criticism of their methodology. Our results question the validity of their whole approach, and specifically indicate that the true predictive capacity of genomes may be higher than their maximal estimates.
Joint work with David Golan.

Saharon Rosset is an Associate Professor in the Department of Statistics at Tel Aviv University. He received his B.Sc. and M.Sc. from Tel Aviv University in Mathematics and Statistics, respectively, and his PhD in Statistics from Stanford University in 2003. From 2003 until 2007 he was a Research Staff Member at IBM Research in New York. He has received grants from the US National Science Foundation, the European Union, the Israeli Science Foundation and IBM. He is an Action Editor of the Journal of Machine Learning Research, and serves on the editorial boards of Machine Learning Journal and Technometrics. He is a four-time winner of the premiere data modeling competition, KDD-Cup. His research interests are in combining statistical, algorithmic and other considerations in developing practical solutions to problems in scientific and business domains.


A Clinician's Perspective on Population Genetics,
Karl Skorecki, Rambam Health Care Campus

Many rare kidney disorders exhibit a monogenic, Mendelian pattern of inheritance. Population-based genetic studies have identified many genetic variants associated with an increased risk of developing common kidney diseases. Strongly associated variants have potential clinical uses as predictive markers and may advance our understanding of disease pathogenesis. These principles are elegantly illustrated by a region within chromosome 22q12 that has a strong association with common forms of kidney disease. Researchers had identified DNA sequence variants in this locus that were highly associated with an increased prevalence of common chronic kidney diseases in people of African ancestry. Initial research concentrated on the gene MYH9 as the most likely candidate gene; however, population-based whole-genome analysis enabled our group and another research team to independently discover more strongly associated mutations in the neighboring APOL1 gene. The powerful evolutionary selection pressure of an infectious pathogen in West Africa favored the spread of APOL1 variants that protect against a lethal form of African sleeping sickness but are highly associated with an increased risk of kidney disease. The presentation will attempt to describe the clinical and epidemiologic background, process of discovery, and reasons for initial misidentification of the candidate gene, remaining challenges, as well as the lessons that can be learned for future population genetics research.

Prof. Skorecki, Director of Medical & Research Development Rambam Health Care Campus, is a clinical nephrologist who has been active in the area of human population genetics research, in the context of genealogy, history, and health. He has also made major contributions in cancer and stem cell research. Dr. Skorecki's interest in population genetics began with a series of collaborative research studies, tracing patrilineal genealogies in the Jewish priesthood, and matrilineal genealogies among Ashkenazi Jews, and used the "signatures" delineated to seek out communities whose Jewish or Near East origins have been lost. Using similar approaches, Skorecki's research team has shown that the Druze population represents a contemporary snapshot of the diversity of Near East populations in antiquity.
His research group has moved to biparental genome-wide analyses of Jewish and non-Jewish communities in health and disease. Combining this approach with principles of evolutionary medicine, and comparative clinical epidemiology, his team identified a genetic locus powerfully associated with common forms of chronic kidney disease and hypertension in certain African heritage populations. Skorecki's activities have been widely recognized in terms of prizes and awards, as well as international public and media interest.


Concluding Remarks,
Moshe Levinger, IBM Research - Haifa

Workshop Organizers