bowtie2 quality scores same for every read

[fvnieuwe]

Dear all,

I am trying to map Illumina 2x126bp exome data (Nimblegen v3) with bowtie2. Subsequently, I want to perform variant detection using Samtools.

I ran into following problem:
- When I do an end-to-end paired-end mapping with bowtie2, the resulting sam file shows a mapping quality (column 5, MAPQ) of 0 for every read.
- When I do a local paired-end mapping with bowtie2, the resulting sam file shows a mapping quality (column 5, MAPQ) of 44 for every read.
- Note: both end-to-end and local mapping of the same reads but as single end reads shows a MAPQ value ranging between 1 and 44 (local) or between 1 and 42 (end-to-end)

Example sam output end-to-end:
bowtie2 --very-sensitive -p 8 --no-mixed --no-discordant -x ./bowtie2_hg19/hg19 -1 F10_GCCAAT_L007_R1.fastq -2 F10_GCCAAT_L007_R2.fastq -S F10_ete.sam --met-file F10_ete.out
HISEQ:126:C0HL2ACXX:7:1101:5336:1952 83 chr19 8841203 0 126M = 8841061 -268 TTAAGTTTTCTCCCTCCAGCCTACTGATTCTAGCTGCAGCCTCATTCCTCTTCTCAAGACACCATCCCAGGAAGCCACCACCAATCAATGATGATTGGCATGTGAGATGAAAATTATTTGCCACCC ####A?9(09.0?88B@A::A>@@>>@A@A>CA::BBBBCCCCA;6(67;7B?7==.)(=)GHFC@=8)<HEBD?700):FHGGFDF?GHGIHHG@IIGF<DH@<H@IGGHF<>FBCABDDDA@@@ AS:i:-4 XN:i:0 XM:i:2 XO:i:0 XG:i:0 NM:i:2 MD:Z:46A31T47 YT:Z:CP
HISEQ:126:C0HL2ACXX:7:1101:5336:1952 163 chr19 8841061 0 126M = 8841203 268 TGAGGCACCACGTCCGGCCGAACCTCAAATTTTTCACTATAGGGAGCAGTGGGTAGCTCTTCATTGGAGTGTGGCCAAGCGTCTCATTCAGAGATGCCAGCTGCTTCTTTGCCATCCCATGCCTTG @<@DFFFDBFBBFBGIIIGIJJJJ<9DH8DFGIIC8BGIEGHII>GIEE7=AA?BDCCCCA@CAA;CA?CDDDD?ACBB?BB?B?@DEECC::A@C4@A@BCC9CCCDDD((4@@A88:@C?::AC AS:i:-2 XN:i:0 XM:i:1 XO:i:0 XG:i:0 NM:i:1 MD:Z:110C15 YT:Z:CP


Example sam output local alignment:
bowtie2 --very-sensitive-local --local -M 5 --score-min G,20,8 -p 8 --no-mixed --no-discordant -x ./bowtie2_hg19/hg19 -1 F10_GCCAAT_L007_R1.fastq -2 F10_GCCAAT_L007_R2.fastq -S F10_local.sam --met-file F10_local.out
HISEQ:126:C0HL2ACXX:7:1101:5336:1952 83 chr19 8841203 44 126M = 8841061 -268 TTAAGTTTTCTCCCTCCAGCCTACTGATTCTAGCTGCAGCCTCATTCCTCTTCTCAAGACACCATCCCAGGAAGCCACCACCAATCAATGATGATTGGCATGTGAGATGAAAATTATTTGCCACCC ####A?9(09.0?88B@A::A>@@>>@A@A>CA::BBBBCCCCA;6(67;7B?7==.)(=)GHFC@=8)<HEBD?700):FHGGFDF?GHGIHHG@IIGF<DH@<H@IGGHF<>FBCABDDDA@@@ AS:i:244 XN:i:0 XM:i:2 XO:i:0 XG:i:0 NM:i:2 MD:Z:46A31T47 YT:Z:CP
HISEQ:126:C0HL2ACXX:7:1101:5336:1952 163 chr19 8841061 44 126M = 8841203 268 TGAGGCACCACGTCCGGCCGAACCTCAAATTTTTCACTATAGGGAGCAGTGGGTAGCTCTTCATTGGAGTGTGGCCAAGCGTCTCATTCAGAGATGCCAGCTGCTTCTTTGCCATCCCATGCCTTG @<@DFFFDBFBBFBGIIIGIJJJJ<9DH8DFGIIC8BGIEGHII>GIEE7=AA?BDCCCCA@CAA;CA?CDDDD?ACBB?BB?B?@DEECC::A@C4@A@BCC9CCCDDD((4@@A88:@C?::AC AS:i:248 XN:i:0 XM:i:1 XO:i:0 XG:i:0 NM:i:1 MD:Z:110C15 YT:Z:CP

Who can make sense of this? Thanks! I already spend a day on this.

Kind regards,
Filip

[fvnieuwe]

I want to bump this question. Mapping paired-end reads with bowtie2 should be common practice. Right? I cannot imagine that nobody elso ran in to this problem. Am I overlooking something really obvious? In the mean time, I helped myself by switching to bwa. I still want to make the bowtie2 approach work, because of speed issues and the local aligment features. Where are the Bowtie aficionados?

[fjrossello]

Hi Filip,

The same problem here (end to end, MapQual all 0s. However, this only happened when --no-mixed --no-discordant options were used. Find the parameters below:

bowtie2 -p 6 --phred33 -N 1 -L 25 -X 650 -I 0 --no-mixed --no-discordant --un-conc ****_un.fastq --al-conc ****_al.fastq --rg-id Flowcell1_Lane4 --rg LB:Lib-**** --rg SM:*** --rg PL:HiSeq2000 -x $bowtie2_reference_dir/genome -1 $rawi/****_1.fq -2 $rawi/****_2.fq -S ****.sam >> ****.out 2>&1

I found no issues when run in --non-mixed mode only. Find commands below:

bowtie2 -p 6 --phred33 -N 1 -L 25 -X 650 -I 0 --no-mixed --rg-id Flowcell1_Lane4 --rg LB:Lib-**** --rg SM:*** --rg PL:HiSeq2000 -x $bowtie2_reference_dir/genome -1 $rawi/****_1.fq -2 $rawi/****_2.fq -S ****.sam >> ****.out 2>&1

I hope this helps.

Cheers,

Fernando

[fvnieuwe]

Thanks Fernando! Indeed, the problem only occurs when using --no-mixed and --no-discordant. I reported this bug to the sourceforge.net site.

[fjrossello]

Excellent! Thanks.
Cheers,
Fernando