Convert BAM files to paired end FASTQ

Question

[berke.toptas]

I have a bunch of bam files aligned with paired-end fastqs and I need to convert them back to paired-end fastqs.

I am using "samtools fastq" for this purpose (after sorting bam files by name):

samtools fastq -1 output.pe_1.fastq -2 output.pe_2.fastq -s singleton.fastq input.bam

The problem is, I noticed that my fastq files have 2 different naming conventions:

1. Case1:

read name in pe1: @UNC15-SN850_90:5:1101:1195:2138/1

read name in pe2: @UNC15-SN850_90:5:1101:1195:2138/2

1. Case2:

read name in pe1: @UNC14-SN744:189:D09V4ACXX:1:1101:1202:1856 1:N:0:TTAGGC

read name in pe2: @UNC14-SN744:189:D09V4ACXX:1:1101:1202:1856 2:N:0:TTAGGC

---

For case 2, I get my paired-end fastq files correctly. However for case 1, all of the reads are pushed to the singleton.fastq and also samtools generate the 2 empty paired-end fastq files.

Is there a way I can smoothly run both cases correctly using "samtools fastq" or any other tools available?

[GenoMax]

Is this TCGA data?

You can may be able to use reformat.sh from BBMap suite if all you need to do is to reformat the /1 id's to ones that contain :.

spaceslash=t            Put a space before the slash in addslash mode.
addcolon=f              Append ' 1:' and ' 2:' to read names, if not already present.

[finswimmer]

Hello,

could you please post complete example lines from the bam files, where one can see the read pairs?

Thanks!

fin swimmer

Answer 1

[JC]

The problem could be Samtools is expecting the second header format (including index). You can try bedtools or Picard for this.