QBiC-Pred: quantitative predictions of transcription factor binding changes due to sequence variants

Area of Science:

• Genomics and Bioinformatics
• Molecular Biology
• Computational Biology

Background:

• Non-coding genetic variants can alter gene regulation by disrupting transcription factor (TF) binding.
• Existing DNA-binding models for TFs often lack quality assessment, hindering the evaluation of mutation impact.
• Accurate prediction of TF binding changes is crucial for understanding variant effects in regulatory regions.

Purpose of the Study:

• To develop a computational tool, QBiC-Pred, for predicting quantitative changes in TF binding affinity due to nucleotide variants.
• To provide statistical confidence (P-values) for predicted TF binding alterations.
• To offer a user-friendly web server for analyzing mutation datasets and their impact on TF binding.

Main Methods:

• Developed QBiC-Pred, a web server utilizing regression models of TF binding specificity.
• Trained models using ordinary least squares (OLS) on high-throughput in vitro binding data.
• Leveraged OLS distributional results to compute P-values for predicted binding changes.

Main Results:

• OLS-based models accurately predict TF binding changes in vitro and in vivo.
• QBiC-Pred outperforms traditional Position Weight Matrix (PWM) models and recent deep learning approaches.
• The web server accepts various mutation dataset formats and provides accessible result post-processing.

Conclusions:

• QBiC-Pred offers a statistically robust method for predicting the functional impact of genetic variants on TF binding.
• The tool enhances the interpretation of non-coding variants by providing reliable predictions and confidence scores.
• QBiC-Pred is a valuable, freely available resource for researchers studying gene regulation and genetic diseases.