deep learning predict gene expression

Question

[BenHu]

how to utilize deep learning to predict (differential) gene expression based on RNAseq, ChIPseq and HiC?

[jamietmorton]

Do you mean predict based on covariates? What is your objective?

[BenHu]

I want to know which epigenetic features, such as promoter, enhancer, or promoter-enhancer interactions, affect gene expression?

[russhh]

All of the above and more.

[BenHu]

yes. But I want to systematically to predict these elements.

[shoujun.gu]

No ml tools can do these tasks now. You may develop one by your own..

Answer 1

[sim.j.baum]

A very interesting topic IMO. I think there is no easy answer to your question because you would need a lot of multi-dimensional data to say that element A is regulating gene B - and you would need functional validation with wet-lab experiments. However, what came to my mind is ChromHMM (http://compbio.mit.edu/ChromHMM/). This characterizes your genomic elements based on ChIP-seq data - however interpretation of the elements is on you. Another paper which i found is this one: https://bmcgenomics.biomedcentral.com/articles/10.1186/s12864-019-5997-2. Hope it helps & best wishes

[BenHu]

thanks. I am reading that paper. best

Answer 2

[i.sudbery]

This is obviously an interesting topic, but it is not a solved question currently. I would investigate the work of Anshul Kundaje, who's group works in this area. https://sites.google.com/site/anshulkundaje/Home