BayesMotif

BayesMotif detects de novo anchored protein sorting motifs using a Bayesian classifier to identify conserved anchor regions and surrounding sequence patterns for protein sorting signal discovery.

Key Features:

• Bayesian classifier: Formulates motif discovery as a classification task and uses a Bayesian classifier-based algorithm to distinguish motif-containing sequences from background.
• Anchored motif model: Identifies motifs composed of a short, highly conserved anchor region paired with less-conserved surrounding regions.
• N- and C-terminal targeting: Targets motifs located at either the N-terminal or C-terminal regions of proteins.
• Impure-dataset robustness: Operates effectively on impure datasets where only a small fraction of sequences (as low as 20%) contain true motif instances.
• Iterative false-positive removal: Uses an iterative procedure to remove sequences unlikely to harbor genuine motifs, refining the dataset and motif signal.
• Meta-sequence feature integration: Facilitates incorporation of meta-sequence features such as hydrophobicity and charge alongside sequence patterns.
• Improved detection over PWM/MEME for certain motifs: Demonstrates superior detection of less-conserved motifs with short conserved anchors compared to conventional PWM-based approaches such as MEME.
• Experimental validation: Validated on implanted motif datasets and real-world protein sequence data.

Scientific Applications:

• De novo motif discovery: Discovery of novel anchored protein sorting motifs in genome-scale protein datasets.
• Targeting signal identification: Identification of N-terminal and C-terminal protein sorting signals that direct subcellular or extracellular localization.
• Noisy dataset analysis: Detection of true motifs in datasets with low motif prevalence or substantial background noise.
• Biochemical characterization of motifs: Characterization of motif properties using meta-sequence features such as hydrophobicity and charge.

Methodology:

Formulates motif discovery as a classification task and employs a Bayesian classifier with an iterative false-positive removal procedure, and supports integration of meta-sequence features such as hydrophobicity or charge.