Prediction and functional interpretation of inter-chromosomal genome architecture from DNA sequence with TwinC

2026-04-20 · Source: Machine learning : nature.com subject feeds · Field: Science & Research — Life Sciences & Biology, Artificial Intelligence & Machine Learning · Depth: Expert, long

Summary

TwinC, an interpretable convolutional neural network, reliably predicts inter-chromosomal (trans) DNA contacts, an under-characterized aspect of 3D genome folding. Current models primarily focus on intra-chromosomal (cis) folding, overlooking these trans-genome interactions. TwinC achieved an AUROC of 0.80 on a cross-chromosomal test set derived from in situ and intact Hi-C experiments in heart tissue. The model was also trained using in situ Hi-C data from the GM12878 cell line and validated with orthogonal DNA SPRITE assays in the same cell type. Mechanistic analysis revealed that TwinC learns the significance of compartments, chromatin accessibility, clustered transcription factor binding, and G-quadruplexes in forming these trans contacts, thereby illuminating their role in gene regulation.

Key takeaway

For AI Scientists and Research Scientists developing 3D genome folding models, you should integrate inter-chromosomal (trans) contact prediction to enhance model completeness and biological relevance. TwinC demonstrates that incorporating features like chromatin accessibility and G-quadruplexes significantly improves predictive accuracy for these previously overlooked interactions, offering a more comprehensive understanding of gene regulation. Consider leveraging the publicly available TwinC code and data for your own research.

Key insights

TwinC predicts inter-chromosomal DNA contacts, revealing their mechanistic drivers and role in gene regulation.

Principles

• Trans-genome organization is crucial for 3D genome folding.
• Compartments and G-quadruplexes influence trans contacts.

Method

TwinC is a convolutional neural network trained on proximity ligation-dependent (Hi-C) and independent (DNA SPRITE) chromatin conformation assays to predict trans contacts.

In practice

• Use TwinC to predict trans contacts from DNA sequence.
• Analyze TwinC's interpretations for regulatory insights.

Topics

• TwinC Model
• Inter-chromosomal Genome Architecture
• Convolutional Neural Networks
• 3D Genome Folding
• Hi-C Assays

Code references

• Noble-Lab/twinc
• Noble-Lab/twinc_paper

Best for: AI Scientist, Research Scientist

Related on AIssential

• A generalizable Hi-C foundation model for chromatin architecture, single-cell and multiomics analysis across species — HiCFoundation integrates 3D chromatin structure and epigenomic data across species using a generalizable foundation model.
• Explicit dynamic cross-strand interactions for DNA sequence language modelling — Explicitly modeling DNA's double-strand interactions improves genomic sequence language model performance and efficiency.
• Why the Human Genome’s Tangled Physicality May Confound AI — The human genome's dynamic, multi-layered regulation defies simple linear models, challenging AI's current sequence-based predictive approaches.
• Two-dimensional geometric template diffusion for boosting single-sequence protein structure prediction — TDFold uses 2D geometric template diffusion for efficient, accurate single-sequence protein structure prediction.
• Machine and Deep Learning Reveal Sequence Determinants Encoding Bivalent Histone Modifications — Distinct DNA sequence features and motifs encode bivalent histone modifications, crucial for stem cell pluripotency.
• $\textit{BlockFormer}$ : Transformer-based inference from interaction maps — BlockFormer infers per-entity parameters from variable interaction maps using synthetic data and a block-aware transformer with 3D positional encoding.

Open in AIssential →

Editorial summary, takeaway, and curation by AIssential. Original article published by Machine learning : nature.com subject feeds.