scaffolding with n= and N=

Question

[jdoyle30724]

Hi!

I have recently run some assemblies with ABySS 2.0.2 and identical parameters with the exception of including or not including N=5 (see below).

abyss-pe lib='pet1 pet2 pet3 pet4 pet5 pet6 pet7 pet8 pet9 pet10 pet11' mp='mp1 mp2 mp3 mp4 mp5 mp6 mp7 mp8 mp9 mp10 mp11' k=51 np=48 j=24 name=k51_N5 n=10 N=5 s=202 l=30 **blah blah libraries**

abyss-pe lib='pet1 pet2 pet3 pet4 pet5 pet6 pet7 pet8 pet9 pet10 pet11' mp='mp1 mp2 mp3 mp4 mp5 mp6 mp7 mp8 mp9 mp10 mp11' k=51 np=48 j=24 name=k51 n=10 s=202 l=30 **blah blah libraries**

My understanding is that for the first assembly 10 reads should be required to build a contig (because n =10) and 5 reads should be required to join contigs into scaffolds (because N=5). For the second assembly 10 reads should be required to build a contig (because n =10) and 10 reads should be required to join contigs into scaffolds (because I haven't specified a different setting for N, so n is the default).

However, when I run these two assemblies, I end up with identical summary statistics (i.e. the exact same number of scaffolds greater than 500 bp in size, same max scaffold length, etc.) It seems like only the n= setting is being applied to the assemblies? Am I somehow specifying the N= parameter setting incorrectly?

Many thanks for your time.

Answer 1

[benv]

Hi @jdoyle30724,

Your description of the behaviour of the n and N parameters is 100% correct. They are the number of paired alignments required to support a link between sequences during the contig-building and scaffold-building stages, respectively. And by default N has the same value as n.

I suspect that the scaffold results just happen to be the same for N=5 and N=10. You could try a more radical change to test it, e.g. N=100.

Answer 2

[jdoyle30724]

Thanks for this! I am really surprised that N=5 didn't at least subtly change the results at multiple kmers. I'll give N=100 at shot.

[h.mon]

please use ADD COMMENT on benv answer instead of answering here.