QBiC-Pred

QBiC-Pred predicts quantitative changes in transcription factor binding caused by nucleotide variants to assess how sequence alterations affect regulatory interactions between transcription factors and genomic target sites.

Key Features:

• Regression-based predictions: Trains regression models on high-throughput in vitro data and uses ordinary least squares (OLS) estimation to quantify effects of DNA sequence variants on transcription factor binding specificity.
• Statistical confidence: Derives distributional results from OLS estimation to compute P-values for each predicted change in TF binding.
• Performance compared to models: Outperforms position weight matrix (PWM) models and recent deep learning models in predicting the impact of mutations on TF binding in vitro and in vivo.

Scientific Applications:

• Non-coding variant interpretation: Assess functional consequences of non-coding genetic variants on transcription factor binding.
• Gene regulation studies: Elucidate how specific mutations influence gene regulation via altered TF–DNA interactions.
• Genetics, genomics and personalized medicine: Prioritize variants that alter regulatory TF binding for studies in genetics, genomics, and personalized medicine.

Methodology:

Regression models are trained on high-throughput in vitro datasets and fitted using ordinary least squares estimation, with P-values computed from the associated OLS distributional results.