TFpredict

TFpredict integrates sequence similarity searching and supervised machine learning to identify and classify transcription factors, determine their structural superclasses, identify DNA-binding domains, and predict cis-acting DNA motifs across eukaryotic genomes.

Key Features:

• Four-Step Classification Workflow: Implements a structured workflow that discriminates TFs from non-TFs using a numeric sequence representation; determines structural superclasses of TFs; identifies DNA-binding domains; and predicts cis-acting DNA motifs.
• Novel Numeric Sequence Representation: Combines BLAST scans with machine learning-based classification into a numeric sequence representation to improve reliability of TF identification and classification.
• Integration of Existing Tools: Extends and adapts existing tools for DNA-binding domain identification and cis-acting DNA motif prediction to provide comprehensive TF analyses.

Scientific Applications:

• Genome-wide TF annotation: Enables genome-wide identification and structural classification of transcription factors in eukaryotic genomes.
• Gene regulatory network reconstruction: Supports inference of TF–cis-regulatory motif associations for exploration of regulatory networks.
• Studies of transcriptional regulation and environmental response: Facilitates research into gene expression regulation and organismal adaptation to changing environmental conditions.
• Comparative and evolutionary analyses: Supports evolutionary and systems biology studies of TF families, structural superclasses, and DNA-binding domains.

Methodology:

The method combines sequence similarity searching (BLAST scans) with supervised machine learning applied to a novel numeric sequence representation to discriminate TFs from non-TFs, assign structural superclasses, identify DNA-binding domains, and predict cis-acting DNA motifs.