problem of mapping SOLiD data using tophat2

[deuterium]

When I using tophat to map SOLiD data (dataset: DRR013130) with the following parameters (tophat2 -m 2 -p 8 --color --quals --bowtie1 -G /home/mm10/mm10_GTF/genes.gtf -o /media/DRR013130_Oo_3_F3 /home/mm10/BowtieCIndex/genome /media/DRR013130_Oo_3_F3.csfasta /media/DRR013130_Oo_3_F3_QV.qual), I found out that each length of the quality of sequence in unmapped.fastq (which is converted from unmapped.bam using bam2fastx) is 1 length shorter than itself length. However, this problem does happen in accepted_hits.fastq (converted from accepted_hits.bam). How could it happen and how to solve it?
Thank you very much for your help!

-----------------------------------------------------------------------------------
>head unmapped.fastq
@DRR013130_Oo_3.3872899
TTGACCTTACGCTCGTGTCAATTGAACTCTTATGCTACCCCTACCGTCAGA 51 length
+
@==;@==?>@?@@@@?@@@@@?;@@2@:/66226?6/8/6?8;/=;;>8/ 50 length
@DRR013130_Oo_3.12296928
TGACCGTCTTAGACATATCTCCGTCGTAGGGATCCCCGGCTAACGGATCCG 51 length
+
@@@@@@@@@@@@@@@@@@@@@@@?@@@;=@@@?6?@@6//2//66//26/ 50 length
-----------------------------------------------------------------------------------
>head accepted_hits.fastq
@DRR013130_Oo_3.9803742
ACTTTTACAAGGCCTAATGGTGACTCCTACAGTGGTTGACACCGACTACC 50 length
+
@__]L@Q__UU]]___[[____]]^^___WW____UU__22___[QS]W8 50 length
@DRR013130_Oo_3.15175672
CAACCTAAAATAAAAACAACTAAAAAAGCTGACTCGTGAGGCAAAAAGAC 50 length
+
@________^^___________]]__________________]]__[[_@ 50 length

[deuterium]

I think I should remove the "T" at the head of each sequence in unmapping.fastq, could anyone help me to write a script to do that?
Thanks very much!

[WhatsOEver]

If the length of the sequence and corresponding quality scores is not equal, you have an error in your input data! (I'm actually a little surprised that tophat is not complaining about these entries) How do you know that the quality score from the leading "T" is missing and not from any other base? Maybe quality score conversion failed totally for these reads (for what ever reason - you have to look into the raw data to check this).

[Brian Bushnell]

Solid data has 1 more "bases" than qualities, because it starts with one fixed base (in this case T) followed by numbers (0-3). But what you are showing is not Solid data. You need to go back to the original colorspace data and map it in colorspace; Solid data cannot be accurately converted to bases without mapping first.

[WhatsOEver]

But the data was mapped in colorspace.

(tophat2 -m 2 -p 8 --color --quals --bowtie1 -G /home/mm10/mm10_GTF/genes.gtf -o /media/DRR013130_Oo_3_F3 /home/mm10/BowtieCIndex/genome /media/DRR013130_Oo_3_F3.csfasta /media/DRR013130_Oo_3_F3_QV.qual

I didn't know that you cannot convert at all without mapping. I would be interested to know how tophat/bowtie is converting the unmappable reads in this case?! If it is really just an additional T at the start you could simply use sed to remove it:

 sed 's/^T//g' ./yourFastqFile.fastq

This will delete 1 leading "T" at the beginning of each line (your quality lines should not have a "T" in them, so there is no need to handle that)

[dpryan]

@WhatsoEver: It's a really bad idea to handle colorspace data in basespace.

[WhatsOEver]

It wasn't my idea, just the answer to the authors question And to point that out: I didn't mean to remove the leading "T" in the ".csfasta" file, I meant to delete it in the "unmapped.fastq" file.

Anyhow, the question would then be of what value is the data in unmapped.bam? If it cannot be used, why/how is it converted?

[deuterium]

@WhatsOEver
Thank you very much! This script is really helpful!

[deuterium]

It wasn't my idea, just the answer to the authors question And to point that out: I didn't mean to remove the leading "T" in the ".csfasta" file, I meant to delete it in the "unmapped.fastq" file.

Anyhow, the question would then be of what value is the data in unmapped.bam? If it cannot be used, why/how is it converted?

Yes, I think it is a bug of tophat2!