High sensitivity TSS prediction: estimates of locations where TSS cannot occur

Area of Science:

• Genomics
• Computational Biology
• Molecular Biology

Background:

• Mammalian genomes contain diverse transcription initiation sites, but most genomic locations are not prone to transcription initiation.
• Identifying non-transcription start sites (NTLs) is crucial for efficient promoter and gene discovery.
• Accurate NTL identification aids in assessing transcript completeness and improving gene annotation.

Purpose of the Study:

• To develop a computational methodology for accurately identifying non-transcription start sites (NTLs) in mammalian genomes.
• To distinguish genomic regions unlikely to harbor transcription start sites (TSSs) from active ones.
• To provide a tool for enhancing gene finding and annotation accuracy.

Main Methods:

• Utilized comprehensive Cap Analysis of Gene Expression (CAGE) and transcript data from mouse and human genomes.
• Developed a high-sensitivity computational TSS prediction algorithm incorporating genomic neighborhood features.
• Applied statistical analyses to features in upstream and downstream regions of known TSSs to define NTL characteristics.

Main Results:

• Successfully annotated large portions of mammalian genomes as NTLs with high accuracy and strand specificity.
• Analysis of human chromosomes 4, 21, and 22 estimated 46%, 41%, and 27% as NTLs, respectively.
• Indicated that over 40% of the human genome is likely to be NTLs.

Conclusions:

• The developed method accurately identifies extensive NTLs in mammalian genomes using high-sensitivity TSS prediction.
• This approach offers a significant advancement for promoter and gene finding, and genomic annotation.
• An implemented server is available for public use at http://cbrc.kaust.edu.sa/ddm/.