Duplicates of transcript in "OUT.read_assignments.tsv"

[P10911004-NPUST]

Hi,

Some of the transcripts have two identical records in the output file ("OUT.read_assignments.tsv"). What is the indication of those duplicates?

#read_id	chr	strand	isoform_id	gene_id	assignment_type	assignment_events	exons	additional_info
transcript/170970	Chr1	-	AT1G01020.1	AT1G01020	unique	fsm,tes_match_precise:-3	8571-8713	PolyA=False;
transcript/170619	Chr1	-	AT1G01020.5	AT1G01020	inconsistent	intron_retention:7836-7941	6833-7069	PolyA=False;
transcript/170619	Chr1	-	AT1G01020.5	AT1G01020	inconsistent	intron_retention:7836-7941	6833-7069	PolyA=False;
transcript/2476	Chr1	+	AT1G01040.1	AT1G01040	inconsistent	tss_match:22	28432-28707	PolyA=False;
transcript/2476	Chr1	+	AT1G01040.1	AT1G01040	inconsistent	tss_match:22	30902-31197	PolyA=False;
transcript/5023	Chr1	+	AT1G01040.1	AT1G01040	unique_minor_difference	ism_3	30410-30715	PolyA=False;
transcript/2503	Chr1	+	AT1G01040.1	AT1G01040	inconsistent	tes_match_precise:0	27099-27281	PolyA=False;

isoquant.log

Thanks for your great effort on this amazing project. This is really helpful. Thank you.

[andrewprzh]

Hi @P10911004-NPUST

For transcript/2476 it looks ok since there are different exon coordinates. It seems there are 2 different alignments reported in the BAM file.
For transcript/170619 it looks odd. Could you check how many alignments correspond to transcript/170619?

You can try running IsoQuant --no_secondary flag.
For transcript classification you can also try SQANTI3.

Best
Andrey

[P10911004-NPUST]

Hi @andrewprzh

I have rerun the whole pipeline, with --no_secondary flag. Duplicates still exist. Am I safe if I just simply ignore those duplicates?
isoquant.log

Some transcripts (each row was rearranged according to the read_id) were shown as below.
exp.txt

Thank you!

[andrewprzh]

@P10911004-NPUST

Now it looks even more odd. For example, transcript/0 is a FSM, but is also reported as ambiguous, which is self-contradictory.
Could you maybe share your initial FASTA or BAM file? Is it possible some sequences have identical ids?

Best
Andrey

[andrewprzh]

@P10911004-NPUST

I think I found and fixed the issue. I need some more time to double check. It will be out in the next release.

Best
Andrey

[andrewprzh]

And by the way, it's safe to ignore duplicated records.

Best
Andrey

[andrewprzh]

Should be fixed in version 3.4