Deep learning of the tissue-regulated splicing code

[cgreene]

Paper needs to be reviewed carefully for relevance:
https://dx.doi.org/10.1093/bioinformatics/btu277

[cgreene]

Biology of the paper:
Aims to identify whether a given exon is going to be present a high, medium, or low abundance in a given tissue from a series of largely genomic features, but also some tissue features (called "cellular context").

Computational aspects:

• Deep architecture.
• Architecture seems very manually defined - biologically-informative features are essentially appended on to certain layers.
• Features also heavily specified (Table S3). Not learned from data.

Not entirely clear to me off the bat how this fits with our overall question: "What would need to be true for deep learning to transform how we categorize, study, and treat individuals to maintain or restore health?" Some diseases may be related to splicing, so perhaps this fits in a discussion there. Methodologically - this work's importance seems to come largely from its early publication.

@YosephBarash - would love to hear your thoughts on this and the other splicing papers regarding how & where they might fit (see issue #2 for potential structure).

[cgreene]

This is an interesting paper. I've labeled it for the 'study' component. It's not receiving more discussion at this point so I've closed it. We're now using 'open' papers only for items undergoing active discussion.