Improving hit @k - Githubissues

broadinstitute / neural-profiling

1 stars 2 forks source link

Improving hit @k #2

Open michaelbornholdt opened 3 years ago

michaelbornholdt commented 3 years ago

@shntnu @niranjchandrasekaran @gwaygenomics

Here is my first version of an analysis of the LINCS data. Please read and give ideas for improvement. https://github.com/broadinstitute/neural-profiling/blob/main/hit_k/lincs_ana.ipynb

[ ] How should the plot look like, do we only care about the first hit?
[ ] How can I improve the randomization to get a more stable
[ ] Should we split up the analysis into MOAs with large number of Compounds and those with less?

michaelbornholdt commented 3 years ago

@AnneCarpenter This is what Shantanu mentioned in the today's checkup meeting

gwaybio commented 3 years ago

from my perspective, that data frame in Cell 9 is all I need (except for L1000 data too). I also like the simplicity of the plot. I don't have any other ideas for improvement

michaelbornholdt commented 3 years ago

@gwaygenomics https://github.com/broadinstitute/neural-profiling/blob/main/hit_k/for_greg.ipynb