search

FunctionLab / selene

a framework for training sequence-level deep learning networks
BSD 3-Clause Clear License
373 stars 88 forks source link

Support for non-classifier models in IntervalSampler and RandomSampler #148

Open laueste opened 4 years ago

laueste commented 4 years ago

Similar to Issue #59 , expanding the other samplers to include support for regression models and other continuous-value-predicting models would allow users to analyze gene expression data, chromatin accessibility data, and many other common and information-rich biological data types. Precedent includes projects such as Basenji, and I think Selene's framework is flexible and powerful enough to add significantly to the field of available options.

kathyxchen commented 4 years ago

Thank you for posting this! This is on our roadmap and we are hoping to be able to look into it in the near future. (Unfortunately, Selene is not currently anyone's main project, and it's a very small group of us contributing to it. We work on improvements when we can though!) Currently, there are a few updates on the way (i.e. improving sampling speed), but I would like to be able to address this next if possible! Let me know if you'd be interested in giving feedback / contributing to this issue in some way - whenever we do work on it, I'd love to get input from interested users.