Paired-end reads group with --read-length do not take R2 length into account

Hi,

I want to group reads based on UMI, alignment position and read length. I’ve paired-end data. So I use this command to group my reads :

python $umitools/group.py -I sorted.bam --paired --read-length --edit-distance-threshold=1 --group-out=groups.tsv -L stats.txt

When I check the groups.tsv file and extract reads from the same group I found that R1 reads have the same UMI, the same alignment position and the same length. But when I check the associated R2 reads they have same UMI and alignment positions but do not have the same length.

Example for a group with 7 reads:

R1 reads

[1] CTAGCGGCCAGGAGAGACCAGATCGGAAGAGCGCGAACTCCCGTC
[2] CTAGCGGCCAGGAGAGACCAGATCGGAAGAGCGCGAACTCCCGTC
[3] CTAGCGGCCAGGAGAGACCAGATCGGAAGAGCGCGAACTCCCGTC
[4] CTAGCGGCCAGGAGAGACCAGATCGGAAGAGCGCGAACTCCCGTC
[5] CTAGCGGCCAGGAGAGACCAGATCGGAAGAGCGCGAACTCCCGTC
[6] CTAGCGGCCAGGAGAGACCAGATCGGAAGAGCGCGTAATGTGCTG
[7] CTAGCGGCCAGGAGAGACCAGATCGGAAGAGCGCGTAATGTGCTG

R2 reads

    aln 
[1] GGTCTCTCCTGGCCGCTAG-------------------------------
[2] GGTCTCTCCTGGCCGCTAGCTGTCTCTTATACACTCTGACGCTGCCGACG
[3] GGTCTCTCCTGGCCGCTAG-------------------------------
[4] GGTCTCTCCTGGCCGCTAG-------------------------------
[5] GGTCTCTCCTGGCCGCTAG-------------------------------
[6] GGTCTCTCCTGGCCGCTAG-------------------------------
[7] GGTCTCTCCTGGCCGCTAG-------------------------------

the second R2 read has a different length. How can I tell UMI-tools to take R2 length also into account.

Thanks

[IanSudbery]

Just to be clear - if you run in --paired we will look for reads that have the same R1 start and the same R2 start. The lengths of R1 and R2 can be different (without --read-length) BUT the length from the start of R1 to the start of R2 will be the same.

[TomSmithCGAT]

I’m closing this issue due to low activity