Predicting the effect of CRISPR-Cas9-based epigenome editing.

Area of Science:

• Genomics
• Epigenetics
• Computational Biology

Background:

• Functional links between epigenetic regulation and mammalian transcription are not fully understood.
• Histone post-translational modifications (PTMs) play a crucial role in orchestrating gene expression.

Purpose of the Study:

• To develop machine learning models for predicting gene expression from histone PTMs.
• To experimentally validate model predictions regarding histone acetylation (H3K27ac) and gene expression.
• To assess model performance in predicting expression changes from engineered epigenetic modifications.

Main Methods:

• Training machine learning models on epigenomic and transcriptomic data from 13 ENCODE cell types.
• Utilizing the dCas9-p300 system for targeted histone acetylation in HEK293T and K562 cell lines.
• Performing MNase-seq to map nucleosome occupancy in HEK293T cells.

Main Results:

• Machine learning models achieved high transcriptome-wide correlations (0.70-0.79) for predicting gene expression from endogenous histone PTMs.
• Models accurately predicted known associations, such as H3K27ac near the transcription start site (TSS) increasing expression.
• Models showed diminished ability to rank fold-changes within individual genes following engineered H3K27ac deposition compared to predicting endogenous expression.

Conclusions:

• Machine learning can effectively predict gene expression from endogenous epigenetic states.
• Predicting expression changes from engineered epigenetic modifications requires further development of epigenome editing tools and causal models.
• Improved datasets and models are crucial for understanding and controlling the human epigenome for health applications.