Paired-end reads : compute insert size along chromosomes

[BurgundiaPR]

Hi everybody,

Does anybody know a tool which take a BAM/SAM in input in order to compute the mean of paired-end insert size along the reference sequences - i.e. human chromosomes ?

If not, what combination of existing tools may give the fastest result ? I planned to use samtools to cut the BAM in little regions, then any tool which compute the mean insert size, but this method will probably be very slow.

Thanks by advance for your advice !
BPR

[GenoMax]

If you want an overall representation of insert sizes in your data then this is one option: http://seqanswers.com/forums/showpos...13&postcount=2

Others:
http://www.genescripts.com/calculating-insert-size.php
https://www.biostars.org/p/94246/

https://gist.github.com/davidliwei/2323462

Samtools view should be fast with sorted/indexed bam files.

[BurgundiaPR]

Thanks for your help.

What I would like is to have this information, but for different intervals. This kind of output should be more clear than my explanation :

Chr Borne_1 Borne_2 Mean_Insert_Size
chr1 1 10000 389
chr1 10001 20000 401
....

BPR

[GenoMax]

Something like this (inspired by one of the links in post above):

$ samtools view sorted_indexed.bam "chr1:1-100" | cut -f 9 | sed 's/^-//' | awk '{ total += $0; count++ } END { print total/count }'

You can then iterate over the intervals.