IceQream: Quantitative chromosome accessibility analysis using physical TF models

Area of Science:

• Genomics
• Computational Biology
• Epigenetics

Background:

• Single-cell sequencing reveals chromosomal accessibility patterns, improving epigenomic prediction from DNA sequence.
• Current quantitative models lack interpretable components for epigenome explanation.
• Predictive modeling of epigenomic activity from sequence is an active research area.

Purpose of the Study:

• To develop a novel modeling strategy and inference algorithm for regressing accessibility from sequences using physical models of transcription factor (TF) binding.
• To create a mechanistic and interpretable baseline for understanding gene and genome regulation from sequence.
• To infer TF effective concentrations and their interactions.

Main Methods:

• IceQream (IQ) employs spatial integration of sequences across TF-DNA affinities and localization relative to target loci.
• IQ infers TF effective concentrations as latent variables, modeling non-linear activation/repression of regulatory elements.
• Synergistic and antagonistic pairwise TF interactions are incorporated.

Main Results:

• IQ demonstrates performance comparable or superior to state-of-the-art deep neural network models on human and mouse data.
• The model successfully infers TF effective concentrations and their interactions.
• IQ provides interpretable components for epigenome modeling.

Conclusions:

• IQ offers a mechanistic and explicable approach to modeling epigenomic activity from sequence.
• This method serves as a valuable baseline for future research in gene regulation.
• IQ advances the understanding of how DNA sequence dictates regulatory element function.