Skip to content

Latest commit

 

History

History
12 lines (12 loc) · 885 Bytes

README.md

File metadata and controls

12 lines (12 loc) · 885 Bytes

Notes

  • A good point for discussion might be that when not enough samples is available it might be better to use a simpler model than a neural network
  • What kinds of gene lengths should be used for normalizing? => lets use mean exon lenght
  • Ensembl ids for rna-seq data are outdated, so retrieving gene lengths is not easy and would require significant amounts of manual annotations
  • Could add one more task for mutations
    • SPL (splicing), EPI (epigenetic), WT (wild type), CTR (control)
  • Could encode the strand as an additional feature, might be especially useful for feature based gnns where each node is a feature
  • Fix up the mRMR package packaging
    • polars broken
    • no checks for whether y is the same length as X
    • no checks for duplicate column names -> results in a long computation and eventually a pandas crash
    • apparently no cicd to update the pkg on pypi