[delasa]

the dexseq_count.py is just yelling at me "claims to have an aligned mate which could not be found. (Is the SAM file properly sorted?)"! I sorted the bam files using "samtools sort" and then converted it to sam file using samtools view but Is still have these messages!

[Simon Anders]

Have you checke your free space. Out of quota maybe? Or maybe 'sort' puts its temporary files to /tmp (see option -T), an at least on our big server, this one is on a small partition which is always full. Also, use '-S 100G' to tell 'sort' that you have lots of memory. BTW, why '-s -c'? We don't need a stable sort.

[Simon Anders]

'samtools sort' sort by position. Use 'samtools sort -n' to sort by name.

[wangxj]

I have the same question as below.

Can anyone answer those questions?

very appreciate

Dear all.

I'm trying to find information about how HTSeq counts reads. I understood that one pair (properly paired) is counted as 1 count.
What about pairs that are not flagged as 'properly paired'?
What about the reads that lost their mate and became single reads?
Are they counted as 1 count as well? Or not counted at all?

Additionally I'm loosing quite many reads that have multiple mappings. Anyone figured out a way to deal with this in HTSeq, instead of just throwing them all out?

[dpryan]

A read that lost its mate will be counted once and a warning will be produced if the unmapped mate isn't actually in the file (tophat does this). I don't recall htseq-count caring about the properly paired flag.

Regarding multimappers, you don't know with certainty where they align, so the proper solution for downstream analyses toward which htseq-count is oriented would be to discard them.

[jkbonfield]

This task is a related one to the bamtofastq conversion in that collation by name is necessary, but not necessarily a full sort.

Collation is often fast, while sorting is very slow. I don't know if there are dedicated collation tools out there (but I'd be suprised if there aren't).

[gt1]

This task is a related one to the bamtofastq conversion in that collation by name is necessary, but not necessarily a full sort.

Collation is often fast, while sorting is very slow. I don't know if there are dedicated collation tools out there (but I'd be suprised if there aren't).

Try bamcollate2 in biobambam (https://github.com/gt1/biobambam).

[Gonza]

Dear All,

Why not sort them BAM files instead of the SAM? SAM takes to much space. After running tophat2, I am thinking something like this:

$samtools sort -n accepted_hits.bam output.bam

then count using htseq-count.

Thoughts????

[dpryan]

There's no need to name sort anymore, htseq-count can handle coordinate sorted BAM files.

[jkbonfield]

Dear All,

Why not sort them BAM files instead of the SAM? SAM takes to much space. After running tophat2, I am thinking something like this:

$samtools sort -n accepted_hits.bam output.bam

then count using htseq-count.

Thoughts????

You will of course get output.bam.bam. Oh how I hate thee Samtools!

[zaki]

Dear all,

There's no need to name sort anymore, htseq-count can handle coordinate sorted BAM files.

Looking at htseq version 0.6.0, the help doc mentioned

 -r ORDER, --order=ORDER
                        'pos' or 'name'. Sorting order of <alignment_file>
                        (default: name). Paired-end sequencing data must be
                        sorted either by [B]position[/B] or by read name, and the
                        sorting order must be specified. Ignored for single-
                        end data.

Please forgive this very naive question, does position sorted bam = coordinated sorted bam? Im guessing its yes, but a conformation would be reassuring.

Many thanks

[dpryan]

Yes, position sorted is another name for coordinate sorted.