TFBIND

TFBIND predicts transcription factor binding sites on DNA to identify regulatory elements and support analysis of transcriptional regulation.

Key Features:

• Positional Weight Matrices (PWMs): Employs PWMs to calculate matching scores between input DNA sequences and known TF binding site frequency matrices using Bucher's calculating approach.
• Cut-off Value Determination: Generalizes the concept of local overrepresentation with statistical methods and uses an iterative algorithm to determine parameters that separate phenomena-dependent from phenomena-independent subsets.
• Background Rate Estimation: Estimates background rates on non-promoter regions to account for potential mis-recognition events and refine cut-off values.
• Data Source and Scope: Uses a dataset of 433 non-redundant vertebrate promoters, including viral promoters from the Eukaryotic Promoter Database (EPD) R.50, and is applied to 205 vertebrate TFs with frequency matrices in TRANSFAC Ver.3.4.
• Re-estimation of Cut-off Values: Re-estimates cut-off values for TFs that may bind in regions preferred by other TFs to improve prediction specificity.

Scientific Applications:

• Genetic Network Analysis: Supports mapping of regulatory interactions by predicting TF binding sites across promoter datasets.
• Gene Function Studies: Aids investigation of gene regulation by identifying potential TF–DNA interactions relevant to gene expression.
• Transcription Regulation Research: Enables study of transcriptional control mechanisms by providing refined TF binding site predictions and cut-off assessments.

Methodology:

PWM-based matching using Bucher's approach, background rate estimation on non-promoter regions, generalization of local overrepresentation via statistical methods, and an iterative algorithm for cut-off value determination and re-estimation.