A comprehensive library of DNA-binding site matrices for 55 proteins applied to the complete Escherichia coli K-12 genome 1 1Edited by R. Ebright

Abstract

A major mode of gene regulation occurs via the binding of speci®c proteins to speci®c DNA sequences. The availability of complete bacterial genome sequences offers an unprecedented opportunity to describe networks of such interactions by correlating existing experimental data with computational predictions. Of the 240 candidate Escherichia coli DNAbinding proteins, about 55 have DNA-binding sites identi®ed by DNA footprinting. We used these sites to construct recognition matrices, which we used to search for additional binding sites in the E. coli genomic sequence. Many of these matrices show a strong preference for non-coding DNA. Discrepancies are identi®ed between matrices derived from natural sites and those derived from SELEX (Systematic Evolution of Ligands by Exponential enrichment) experiments. We have constructed a database of these proteins and binding sites, called DPInteract (available at

A Biophysical Approach to Transcription Factor Binding Site Discovery

“…In the practical, bioinformatics context, the principal difference between the QPMEME and the standard information-based weight matrix method is that the latter does not produce a natural threshold for the "information score" used to classify sequence segments (Robison et al 1998). Hence, search results are often quoted for the high and the low settings of threshold determined, respectively, by the average information score in the example set and the average minus two standard deviations (Robison et al 1998).…”

“…Instead, we are trying to minimize the probability that any random sequence is identified as a binding sequence, while still correctly classifying all of the examples. number of candidate sites found using information theoretic weight matrix with threshold set to include most of the example sites (the 2 cut-off; Robison et al 1998). e number of candidate sites found using QPMEME approach.…”

“…We have used the Quadratic Programming Method of Energy Matrix Estimation (QPMEME) to construct matrices for the DNA-binding proteins ( (tf) ) from the known binding sites (Robison et al 1998) collected in the DPInteract database. We then searched both strands of the 3121 non-ORF regions (with length >10 bp) of the E. coli genome (Blattner et al 1997) to identify the candidate transcription factor binding sites, as defined by the E (tf) (S) = (tf) · S Յ µ = ‫1מ‬ condition.…”

“…The QPMEME algorithm allows identification of the likely RNA polymerase (RNAP) binding sites (promoters) including the RNAP-70 sites. Identification of these sites is highly nontrivial because of their strong variability in E. coli (Robison et al 1998;Wagner 2000). This, in turn, makes it possible-by examining the location of TF-binding sites relative to promoters and to other TF-binding sites (Gralla and Collado-Vides 1996)-to predict the functional modality (repressor vs. activator) for at least some of the candidate regulatory sites.…”

Two types of localization of the DNA‐binding proteins within the Escherichia coli nucleoid

“…Escherichia coli contains at least 240 different molecular species of the DNA-binding protein (Robison et al 1998). A group of several abundant proteins constitutes the compact nucleoid architecture by condensing the genome DNA, which is about 1000-fold longer than…”