Remove paired end duplicated reads -Samtools-Picard

[paolo.kunder]

Dear All,

I am facing some big problem with samtools rmdup command,

What I would like to do is to discard reads that fail to align as proper pairs and remove non-unique fragments (and not non-unique reads), in order to delete paired reads that have both F and R reads starting and ending exactly in the same place.

I tried to run:

to select only properly paired reads

samtools view -f 2 -b -h -o output.bam input.bam

to remove duplicates:

samtools rmdup  input.bam output.bam

but it seems thta reads are not properly removed,

I tried also to run samtools fixmate before running rmdup but even strangest results.

So I Switched to Piacard tools Markduplicates command:

java -jar MarkDuplicates.jar I=input.bam O=output.bam M=marked_rep
ort.txt VALIDATION_STRINGENCY=SILENT REMOVE_DUPLICATES=true

But even if my file is sorted the program fails reporting bam not sorted:

Exception in thread "main" net.sf.picard.PicardException: Input file input.bam is not coordinate sorted.

Any suggestion??

Thanks,
Paolo

[swbarnes2]

Picard isn't actually checking to see if the file is sorted, it's just looking for a line in the header saying it is, but samtools sort will not put that header there.

Use Picard SortSam to sort, and it will adjust the header to say that it is sorted, then MarkDuplicates will work.

[Chipper]

Just tell Picard to assume it is a sorted file (AS=true)