kmer-SVM: a web server for identifying predictive regulatory sequence features in genomic data sets.

Area of Science:

• Genomics
• Computational Biology
• Bioinformatics

Background:

• Massively parallel sequencing generates large genomic datasets, like Chromatin immunoprecipitation followed by sequence assays and DNase-seq, to identify regulatory DNA regions.
• Interpreting these datasets requires identifying the DNA sequence code that defines transcription factor (TF) binding and open chromatin regions.

Purpose of the Study:

• To develop and implement a web server for the kmer-support vector machine (SVM) computational methodology.
• To enable the broader research community to analyze and interpret their genomic datasets using kmer-SVM.

Main Methods:

• Developed a novel computational methodology using a support vector machine (SVM) with kmer sequence features (kmer-SVM).
• Implemented kmer-SVM as a publicly accessible web server.
• Applied kmer-SVM to analyze five recently published genomic datasets.

Main Results:

• The kmer-SVM tool successfully identified predictive combinations of short transcription factor-binding sites.
• Demonstrated the tool's ability to identify accessory factors and repressive sequence elements in analyzed datasets.
• The method provides confidence in genomic experiments by recovering known binding sites and revealing novel sequence features.

Conclusions:

• The kmer-SVM web server provides a valuable resource for the research community to interpret complex genomic data.
• This tool facilitates the discovery of regulatory DNA sequence codes, aiding in the understanding of gene regulation.
• Enables experimental testing of cooperative mechanisms by revealing novel sequence features.