Lab Director


Selected Publications

Prof. Simon Kasif current research area is Bioinformatics, Computational Genomics and Molecular Engineering.  He is currently a Professor of Bioinformatics and Biomedical Engineering at Boston University. He is also affiliated with the Center for Integrating Informormatics and Bedside (I2B2) and Children's Hospital, Harvard Medical School

Previously he held faculty and research positions at Johns Hopkins University, Princeton University, University of Illinois, Chicago, University of Chicago, Cambridge Research Laboratory, NEC Research Institute, HP Labs and MIT Genome Center.

Prof. Kasif was previously working in Artificial Intelligence, Parallel and Distributed Computing, Machine Learning and Algorithmics.   

In Bioinformatics Prof. Kasif has contributed to several areas in bioinformatics and computational biology including:

In 1992 Prof. Kasif and his students at Johns Hopkins pioneered the use of Bayes networks for modeling biological data. The same paper described one of the earliest deployments of Hidden Markov models for structural genomics (ISMB 1993), (secondary structure prediction). Bayesian networks generalize HMMs for analysis of biological data and has been recently used in Kasif's lab and others labs for integrating diverse biological databases, functional genomics and comparative genomics .

Helped develop Glimmer, a widely used microbial DNA gene discovery system.  Glimmer has been used for the analysis of many of the microbial genomes available today. Glimmer was implemented by Art Delcher  (jointly with  Steven Salzberg, Art Delcher and Owen White).

Helped develop Mummer,a system for large scale genomic comparison and SNP detection, Mummer has been used in an effort to identify the minimal set of genes essential for life (Science, 10/12/2000), detecting chromosomal duplications in Arabodopsis (Nature 12/2000) and detecting important inversions and duplications rearrangements in bacterial genomes. was implemented by Art Delcher (joint effort with   Art Delcher and Steven Salzberg).

POMP, a novel multiplex PCR method used to close gaps in several whole genomes including Streptococcus. (jointly with  Herve Tettelin and Steven Salzberg).

A book that describes advanced computational biology and bioinformatics methods.

A decision tree system OC1 that has been used for analysis of scientific data. The system is used in the eukaryotic gene finding system Morgan and more recently for functional genomics. OC1 was implemented by S. Murthy (joint project with Steven Salzberg) 

Prof. Kasif and his students experimented with one of the earliest deployments of Hidden Markov models for for protein structure analysis (ISMB 1993), (secondary structure prediction)

The lab developed several algorithms and systems for functional genomics, structural genomics and comparative genomics.

Prof. Kasif and and Prof. Stan Letovsky popularized one of the first methods for probabilistic functional inference (ISMB 2003)

In 1999, Prof. Kasif was a recipient of NSF's competitive KDI Award for work in Bioinformatics -- "Intelligent Genomic Analysis", $1,900,000 (jointly with TIGR).

The lab now focuses on gene function, computational analysis of molecular processes deregulated in disease, reverse engineering of biological systems, pathway analysis, comparative genomics and new biotechnology development

He also studied artificial intelligence, parallel complexity and algorithms, constraint systems, computational learning theory, cognitive modeling  and biologically insprired computing.


Selected previous computer science projects developed by Kasif and his collaborators are listed below:

First parallel deductive database (logic programming system) to be mapped to a real multiprocessor (ZMOB). Joint work with Madhur Kohli, Rich Piazza and Jack Minker.

The longest chess game: Discovery of chess endgames that require 221 moves to capture a piece and subsequently win. This work was done (entirely independently) by Lewis Stiller as part of his PhD thesis using a very clever dynamic programming formulation and symmetry group representation programmed on a 65000 processor Connection machine.

Relaxation in Constraint Satisfaction Networks is inherently sequential, despite previously held belief it was highly parallelizable.

First log-time algorithms for Bayes networks -- with Judea Pearl, Adam Grove and Art Delcher

A formal model of learning with limited memory. (with Rao Kosaraju Steven Salzberg and David Heath).

Combining supervised (nearest neighbour and unsupervised learning (bayes nets) with David Waltz, Steven Salzberg, John Rachlin, and David Aha. .



Last modified