-
Biology Applications for Computational & Mathematical
Methods - Wix Small Hall
- Eytan Domany, Chair
-
Yoav Benjamini, Department of Statistics and
Operations Research, Tel Aviv University
-
Controlling the False Discovery Rate in Behavioral Genetics and
Microarrays Analysis
Any study based on statistical evidence is prone to produce false
discoveries. Testing statistical hypotheses at the traditional 0.05
level is a way to limit the probability of producing such a false
discovery when it is not real. Alas when many such tests are
conducted in a study, the probability of producing a false discovery
increases dramatically.
On the other hand, limiting this probability in the traditional way
incurs deterioration in the probability of detecting true discoveries.
The control of the false discovery rate (FDR) has been suggested as an
intermediate approach to the problem. Procedures that control the FDR
at a desired level are gaining popularity in areas of science where
the problems encountered are large.
Two such areas will be discussed in this talk behavioral genetics
and gene expression microarray data.
The two will serve as a vehicle to discuss the FDR approach, the simple
procedures that may be used to control it, and their properties.
-
Nati Linial, Department of Computer Science,
Hebrew University of Jerusalem
-
Large-scale Clustering of Protein Sequences: Some Theory and Some
Practice
In this presentation, I will review some of our recent
work on the large-scale classification and analysis
of proteins. I will explain some of the algorithmic
concepts that underlie our work. Among the algorithmic
tools that we employ are: Spectral data
analysis, combinatorial algorithms to detect
significant domains and small distortion
metric embeddings.
-
Gad Getz, Department of Physics and Complex
Systems, Weizmann Institute of Science
-
Cluster Analysis of Gene Expression Data
A single microarray experiment allows simultaneous
measurement of the expression level of thousands of genes. A
typical experiment uses a few tens of such microarrays, each
focusing on one sample - such as material extracted from a
particular tumor. Hence the results of such an experiment contain
several hundred thousand numbers, that come in the form of a
table, of several thousand rows (one for each gene) and 50 - 100
columns (one for each sample).
We developed and applied a "data mining" method, called Coupled Two-Way
Clustering} (CTWC), to extract biologically
relevant data from such matrices. By an iterative clustering
procedure the method reveals correlations that involve small
subsets of genes and samples. The method can identify sets of
correlated genes which usually belong to the same biological
process or pathway, and uncover types and sub-types of diseases.
The algorithm is designed to link cellular conditions to their
relevant genes by finding conditional correlation among them. I
will present results obtained from analyses of several types of
cancer.
The CTWC method was applied by others and by us to numerous data
types including gene expression data, antigen reactivity data,
analysis of sugar compounds, document categorization and
low-temperature phases of short range spin glasses.
