Questions About The Asm-File Produced By The Velvet Program.

Question

[Dane]

Hey all,

To preface, for the project I'm involved with I need to extract paired end info from an assembly - that is, for each paired end read: which contigs it connects, and the orientation and distance it suggests.

To this end, I've assembled on Velvet, created an AMOS file, and then created files for each individual contig from this (using the asmbly_splitter script). However, I'm having trouble interpreting the format of the file, and I've found very little documentation.

For example, the file output looks like:

......
{RED
iid:5769434
eid:5769434
seq:
TACGGGAAATCAGCGGAGGTGATTCCCTTTCAGGGGAAAGAGTGGGTACATCTGAAGGAG
TGACACAGAGTGACAGAGCTGGAAACACATTTGC
.
qlt:
TACGGGAAATCAGCGGAGGTGATTCCCTTTCAGGGGAAAGAGTGGGTACATCTGAAGGAG
TGACACAGAGTGACAGAGCTGGAAACACATTTGC
.
}
......

{TLE
src:5769434
off:327
clr:0,94
}
......
{TLE
src:557270
off:418
clr:0,100
}

Can anyone help me understand what these results represent? And how to extract the information I need from this?

Thanks

Answer 1

[Jeremy Leipzig]

from your wording "contigs it connects", it sounds like you really want to know which pairs are allowing contigs to be scaffolded

you basically want to isolate tiles with mates that appear in two different CTGs (contigs)

the weird thing is that contigs in scaffolds SCF are designated by TLEs

outside of scaffolds they are simply CTGs and the read segments making them up are TLEs

awesome stuff i know

I would start by finding SCF blocks and work your way to finding SRC's that belong to the same FRG (fragment, I can't remember if velvet produces FRG tags - they tie paired ends together). If not then you will have to rely on consecutive reads being pairs like 5769434 and 5769435.

Getting Number Of Reads In A Contig From Velvet Assembly

[Dane]

Thanks. But I still have puzzles about the asm file. I don't know the meaning of element 'clr' in the TLE, and many others. Could you please tell me where to find the exact instruction about the asm file. Actually, I want to finish the edges bundling on the Contigs, and I followed the steps listed in the 'Opera' paper and others. But as a new research in bio-information, it's too difficulty for me, do you know any better or simple methods to do it?