Decide on minimum read count for exon expression

[sr320]

20x

[kubu4]

Is this per exon or mean across all exons per gene ?

[sr320]

I think we want to keep exon with low / no expression. .. Lets set a threshold as sum of read counts for first 6 exons (as this is what we are looking at) to be 1000.

[kubu4]

Okay, when I do this (sum read coverage across first 6 exons per gene), I end up with only 2,497 genes having a sum of >= 1000.

< 1000      >= 1000 
  35763        2501

[kubu4]

Can check out code here:

https://github.com/sr320/ceabigr/blob/a6a5f241329e6a0fd9a647ece32930832a5ccb91/code/65-exon-coverage.qmd#L357-L375

[sr320]

what would the gene count be if reduced sum to 100.

[kubu4]

< 100    >= 100 
  6510    31754

[sr320]

Greater than 500?

[kubu4]

 < 500     >= 500 
 31627       6637

[sr320]

lets go forward with > 100

[kubu4]

Alrighty, we may need to make further adjustments. Those numbers above were just from a single sample that I was using for code testing.

I've managed to write code to look at all the files and do the threshold filtering for all samples on a per gene basis. I.e. All samples must have an exon coverage sum threshold of n.

Threshold	Genes
10	23101
25	18119
50	13827
75	10357
100	7485
500	385

[kubu4]

Relevant code section:

https://github.com/sr320/ceabigr/blob/6b4bf89801a50180993331a927830d820b42e5e3/code/65-exon-coverage.qmd#L437-L486

[sr320]

lets go with threshold of 10