Assessing transcription factor motif drift from noisy decoy sequences

Area of Science:

• Genomics
• Molecular Biology
• Bioinformatics

Background:

• Genome-wide identification of transcription factor binding sites (TFBS) is crucial for understanding mRNA expression regulation.
• Experimental methods identify TF-gene associations but lack precision in pinpointing binding sites.
• Computational methods offer a more general approach to TFBS identification, independent of experimental conditions.

Purpose of the Study:

• To evaluate the robustness of computational TFBS identification methods.
• To address challenges posed by heterogeneous TF binding affinities and the presence of non-binding sequences.
• To propose a novel method for measuring deviations from canonical TF binding sites.

Main Methods:

• Assessing the impact of including non-binding upstream regions on TFBS prediction specificity.
• Developing a method to calculate distances between position weight matrices (PWMs).
• Quantifying "drift" from canonical binding sites using PWM distance.

Main Results:

• The addition of upstream regions lacking TFBS significantly reduces prediction specificity.
• A new method for calculating PWM distances was proposed and evaluated.
• The proposed method effectively measures the "drift" from canonical binding sites.

Conclusions:

• Computational TFBS identification is sensitive to the inclusion of irrelevant sequences.
• The developed PWM distance metric provides a valuable tool for assessing binding site variability.
• Findings can inform the development of more accurate TFBS identification algorithms.