Skip to content

Latest commit

 

History

History
34 lines (24 loc) · 1.1 KB

README.md

File metadata and controls

34 lines (24 loc) · 1.1 KB
Raw

Number of samples

total_samples = 1477 eff_samples = 1449 #no kidney eff_samples = 1339 #no kidney liver

Prefiltering: delete splicing events (SE) with >=0.1 NA's

  • psi.tissue: Merge PSI + description (by sampleId) *NOT in GITHUB

  • notNA_tissue.tsv: percentage of NA per SE and tissue (not kidney) *with dcast() -> delete NA if percentage >= 0.1

  • depurated_notNA_tissue.tsv: percentage of NA per SE and tissue, after deleting SE rows that percentage >=0.1 (only keeping SE <=0.1 NAs).

  • psi.tissue.filtered (sample_id, splicing_event, PSI, tissue): filtered of NAs and no kidney *NOT in GITHUB

(*) graphics:

  • notNA_tissue vs. depurated_notNA_tissue
  • psi.tissue vs. psi.tissue.filtered

Training and Testing sets

training_set = 100/tissue * 5 = 34.50656%

  • training_set = 100 samples/tissue -> 500 samples *NOT in GITHUB

(500/1449)*100 34.50656

  • testing_set = 1449-500 = 949 *NOT in GITHUB

(900/1449)*100 62.1118

To have all_results: concatenated results output from Naive_Bayes_def.py (without header)

Sorted accuracies: $ sort -rnk 1 accuracy_1.out > accuracy_1_sorted.out