chung-lab / SCAFE

Single Cell Analysis of Five'Ends
MIT License
45 stars 11 forks source link

demo output of calculateDirectionality reveals mis-classified directionality? #34

Open mkarikom opened 5 months ago

mkarikom commented 5 months ago

the subroutine calculateDirectionality adds a 'score' to the output file:

SCAFE/demo/output/sc.solo/directionality/demo/bed/demo.CRE.directionality.summit_center.bed.gz

According to the sub, we have that

  1. $strand_diff = abs($fwd_count-$rev_count);
  2. the score column of the above bed file contains $directionality = 1-($strand_diff/$strand_sum);

In other words, if $fwd_count==$rev_countthen score=1 in the bed file.

Running the demo, there is exactly 1 locus with bi-directional transcription (strand==".") and about 30 loci where the score column of the bed file is > 0.9 denoting a strand count imbalance of less than 10%. Where 5 of these loci surprisingly have a score of 1, which should mean completely balanced fwd and reverse strand counts. Such transcriptional activity seems intuitively bidirectional:

image