UMI-tools: reads with different sizes in the same UMI group

Question

[Nicolas Rosewick]

EDIT : Just saw that --read-length parameter answer my question ! My bad...

Hi,

I run UMI-tools (https://github.com/CGATOxford/UMI-tools) on my dataset using this command :

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

After that I've a group.tsv file containing the groups for my reads. Here's a subset of my group.tsv :

read_id contig  position    umi umi_count   final_umi   final_umi_count unique_id
NS500186:291:HTYKYAFXX:3:11511:4602:8498_CTCTTCCGATCT   chr1    504 CTCTTCCGATCT    1   CTCTTCCGATCT    1   0
NS500186:291:HTYKYAFXX:3:21407:18031:16780_CTCTTCCGATCT chr1    504 CTCTTCCGATCT    1   CTCTTCCGATCT    1   1
NS500186:291:HTYKYAFXX:1:11103:22901:14851_CTCTTCCGATCT chr1    504 CTCTTCCGATCT    24  CTCTTCCGATCT    24  2
NS500186:291:HTYKYAFXX:1:11104:5947:6879_CTCTTCCGATCT   chr1    504 CTCTTCCGATCT    24  CTCTTCCGATCT    24  2
NS500186:291:HTYKYAFXX:1:11105:22350:7050_CTCTTCCGATCT  chr1    504 CTCTTCCGATCT    24  CTCTTCCGATCT    24  2
NS500186:291:HTYKYAFXX:1:11207:26170:12155_CTCTTCCGATCT chr1    504 CTCTTCCGATCT    24  CTCTTCCGATCT    24  2
NS500186:291:HTYKYAFXX:1:11209:5982:16287_CTCTTCCGATCT  chr1    504 CTCTTCCGATCT    24  CTCTTCCGATCT    24  2
NS500186:291:HTYKYAFXX:1:11304:14850:18144_CTCTTCCGATCT chr1    504 CTCTTCCGATCT    24  CTCTTCCGATCT    24  2
NS500186:291:HTYKYAFXX:1:11307:16408:8652_CTCTTCCGATCT  chr1    504 CTCTTCCGATCT    24  CTCTTCCGATCT    24  2

So each read harboring the same alignment position and the same umi (with max 1bp distance) are assigned to the same group.

But looking at the reads I've a lot of cases where the read sequence is not the same (in size). Alignment starts is ok but alignment end is not the same because the input read has a different length.

Example, let's take the umi group 13 containing 8 reads :

read_seq    position    umi umi_count   final_umi   final_umi_count unique_id
CTAGCGGCCAGGAGAGACCGCAGGCAGACCGCTTCCCTCCAGGAAGAGCGCCAGTTTTCTAG  8693    CTCTTCCTGGAG    8   CTCTTCCTGGAG    8   13
CTAGCGGCCAGGAGAGACCGCAGGCAGACCGCTTCCCTCCAGGAAGAGCGCCAGTTTTCTAG  8693    CTCTTCCTGGAG    8   CTCTTCCTGGAG    8   13
CTAGCGGCCAGGAGAGACCGCAGGCAGACCGCTTCCCTCCAGGAAGAGCGCCAGTTTTCTAGCTGGCTCTTATAC 8693    CTCTTCCTGGAG    8   CTCTTCCTGGAG    8   13
CTAGCGGCCAGGAGAGACCGCAGGCAGACCGCTTCCCTCCAGGAAGAGCGCCAGTTTTCTAG  8693    CTCTTCCTGGAG    8   CTCTTCCTGGAG    8   13
CTAGCGGCCAGGAGAGACCGCAGGCAGACCGCTTCCCTCCAGGAAGAGCGCCAGTTTTCTAG  8693    CTCTTCCTGGAG    8   CTCTTCCTGGAG    8   13
CTAGCGGCCAGGAGAGACCGCAGGCAGACCGCTTCCCTCCAGGAAGAGCGCCAGTTTAATAG  8693    CTCTTCCTGGAG    8   CTCTTCCTGGAG    8   13
CTAGCGGCCAGGAGAGACCGCAGGCAGACCGCTTCCCTCCAGGAAGAGCGCCAGTTTTCTAG  8693    CTCTTCCTGGAG    8   CTCTTCCTGGAG    8   13
CTAGCGGCCAGGAGAGACCGCAGGCAGACCGCTTCCCTCCAGGAAGAGCGCCAGTTTTCTAG  8693    CTCTTCCTGGAG    8   CTCTTCCTGGAG    8   13

So same alignment position and same UMI. ok. But as you can see one read has a different size (read 3 has 75bp compared to 63 for the others). So for me read3 should not be part of this group. Is it possible to force UMI-tools to only group reads with the same size ?

CTAGCGGCCAGGAGAGACCGCAGGCAGACCGCTTCCCTCCAGGAAGAGCGCCAGTTTTCTAG-------------
CTAGCGGCCAGGAGAGACCGCAGGCAGACCGCTTCCCTCCAGGAAGAGCGCCAGTTTTCTAG-------------
CTAGCGGCCAGGAGAGACCGCAGGCAGACCGCTTCCCTCCAGGAAGAGCGCCAGTTTTCTAGCTGGCTCTTATAC
CTAGCGGCCAGGAGAGACCGCAGGCAGACCGCTTCCCTCCAGGAAGAGCGCCAGTTTTCTAG-------------
CTAGCGGCCAGGAGAGACCGCAGGCAGACCGCTTCCCTCCAGGAAGAGCGCCAGTTTTCTAG-------------
CTAGCGGCCAGGAGAGACCGCAGGCAGACCGCTTCCCTCCAGGAAGAGCGCCAGTTTTCTAG-------------
CTAGCGGCCAGGAGAGACCGCAGGCAGACCGCTTCCCTCCAGGAAGAGCGCCAGTTTTCTAG-------------
CTAGCGGCCAGGAGAGACCGCAGGCAGACCGCTTCCCTCCAGGAAGAGCGCCAGTTTAATAG-------------
*********************************************************  ***

Thanks

[GenoMax]

Add that mistake as an answer and then accept it. That way this thread will not remain open.

Answer 1

[Nicolas Rosewick]

Just saw that --read-length parameter answer my question ! My bad...

--read-length         use read length in addition to position and UMIto
                      identify possible duplicates [default=False]