Distribution patterns of over-represented k-mers in non-coding yeast DNA

Area of Science:

• Genomics
• Bioinformatics
• Computational Biology

Background:

• Over-represented k-mers in genomic DNA can indicate biologically significant regions.
• These k-mers are often associated with transcription factor binding sites in co-regulated gene families.
• Understanding k-mer distribution is crucial for deciphering regulatory elements.

Purpose of the Study:

• To introduce a statistical background model for measuring k-mer over-representation.
• To investigate the context and spatial distribution of over-represented k-mers in yeast upstream regions.
• To relate k-mer distribution patterns to DNA structure, function, and evolution.

Main Methods:

• Development of a statistical background model based on single-mismatches.
• Application of the model to pooled 500 bp ORF Upstream Regions (USRs) of yeast.
• Analysis of spatial distributions, homology, localization, and DNA structure of over-represented k-mers.

Main Results:

• Most over-represented k-mers exhibit non-random spatial distributions, clustering into distinct classes.
• Three common patterns were identified: localized distributions in homologous ORF clusters, broad symmetric distributions for regulatory elements, and broad distributions for A/T-rich runs.
• These patterns correlate with DNA structure, function, and evolutionary relationships.

Conclusions:

• The spatial distribution of over-represented k-mers provides valuable information about their biological roles.
• A data-mining approach integrating over-representation, homology, localization, and DNA structure aids in identifying important k-mers.
• This study contributes to understanding the 'lexicon' of regulatory regions in genomic DNA.