Specific modelling of regulatory units in DNA sequences

Area of Science:

• Genomics
• Bioinformatics
• Molecular Biology

Background:

• Transcriptional control regions feature complex arrangements of individual elements, such as protein binding sites.
• This modular structure enables vast functional diversity in regulatory regions using a limited set of elements.

Purpose of the Study:

• To develop algorithms for creating complex models of transcriptional control regions based on element composition and organization.
• To enable efficient analysis of large genomic sequences for regulatory elements.

Main Methods:

• Implementation of ModelGenerator, an algorithm that uses formal representations of regulatory region features.
• ModelGenerator requires a training set of sequences and an initial model (e.g., two transcription factor binding sites).
• ModelInspector scans new sequence data for matches to models generated by ModelGenerator.

Main Results:

• ModelGenerator successfully expands initial models into complex, biologically meaningful representations through comparative sequence analysis.
• Developed models for retroviral transcriptional control regions demonstrate high specificity.
• ModelInspector exhibits high accuracy, with searches against GenBank showing no false negatives and minimal false positives (<2/million nucleotides).

Conclusions:

• The developed algorithms (ModelGenerator and ModelInspector) are effective tools for analyzing transcriptional regulatory regions.
• These tools are particularly valuable for the analysis of extensive genomic sequences generated by current sequencing projects.
• The approach facilitates a deeper understanding of gene regulation and genomic sequence function.