TIN calculation considers strandedness?

[hy09]

Hi, I was wondering if the TIN score calculation considers the strandedness of reads? I'm looking at Pacbio long read rnaseq data. I noticed some transcripts that got a low TIN do have even coverage but overlap with another transcript on the opposite strand. If strandedness was not taken into account that might explain why they are getting a low TIN.

Here is an example:
DDIT3 has a TIN of 35%, the last exon overlaps with the first exon of MARS1

[jonperdomo]

Hi @hy09 it does not currently calculate strand-specific coverage for the TIN, but I will add this update. Thank you for identifying this.

[hy09]

Awesome thank you! Looking forward to the update.

[jonperdomo]

I have updated this and tested with simulated reads and on two overlapping transcripts.
When I remove the reverse strand reads (blue), the overlapping transcript's TIN (forward strand) is now unaffected:

Test BED file of overlapping transcripts:

chr21	5499154	5502497	ENST00000624859.3	0	-	5502497	5502497	0	3	813,35,129,	0,912,3214,
chr21	5500997	5504340	ENST00000623449.3	0	+	5502497	5502497	0	3	813,35,129,	0,912,3214,

TIN with both forward and reverse alignments:

geneID	chrom	tx_start	tx_end	TIN
ENST00000624859.3	chr21	5499154	5502497	78.72352783773570
ENST00000623449.3	chr21	5500997	5504340	76.25804518209634

TIN with only forward alignments:

geneID	chrom	tx_start	tx_end	TIN
ENST00000624859.3	chr21	5499154	5502497	0.00000000000000
ENST00000623449.3	chr21	5500997	5504340	76.25804518209634