The problem
To model transcriptional regulation, researchers need accurate representations of how transcription factors recognize DNA sequences. Different experiments and computational methods can produce conflicting position weight matrices (PWMs) for the same yeast transcription factor, with varying ability to predict actual occupancy.
The ScerTF approach
ScerTF integrated matrices from eleven scientific sources and systematically compared their predictions against in-vivo chromatin immunoprecipitation (ChIP-chip) data and transcription factor deletion experiments. Each TF was assigned a recommended matrix based on predictive performance and supporting evidence.
Where experiments permitted, the original analysis reported Fisher’s exact-test P-values, matrix information content, and optimized matrix score cutoffs. These measurements, together with candidate matrix comparisons, remain available in each factor’s record.
Finding binding motifs
The archive supports three ways to query the original data: transcription factor names and gene aliases; short DNA sequences matched against recommended motifs; and pasted nucleotide frequency, count, or weight matrices. The browser-based motif search is a port of the original ScerTF alignment and normal-approximation scoring approach, with multiple-testing correction.
Sequence/matrix search results are computational motif matches, not direct proof of transcription factor occupancy in a cellular context.
The scientific publication
Spivak AT, Stormo GD. “ScerTF: a comprehensive database of benchmarked position weight matrices for Saccharomyces species.” Nucleic Acids Research 40 (Database issue), D162–D168 (2012).
Read the primary research paper ↗ · PubMed (PMID: 22140105) ↗
This is a restored archive of the original 2012 data. Interpret descriptions, annotations and cross-references in the context of their publication date.