abyss-sealer

161 views
Skip to first unread message

Nisha pillai

unread,
Aug 1, 2016, 4:09:44 AM8/1/16
to ABySS, Esakimuthu Pillai Nisha PILLAI (IMCB)

Hello there,

I am new to abyss 'sealer'. Trying it out on our new assembly.

Couldn't figure out the following from the sealer BMC bioinformatics paper. So posting it here.

(1) In this assembly we don't know the exact size/number of the gaps. This assembly has a representative 100 Ns for every gap.
     
     Does it treat the number of Ns as the exact gap or I am wondering how does Sealer handle this? 


(2) In-order to understand the gap distribution in Human reference, I downloaded GRCh38 (based on the sealer 'BMC bioinformatics' paper).
 
     I find that the total gaps here is only 1242; while in the paper it is 237,408. 

     So can you advise me which is the reference that was used in the paper ?


Many thanks!
Kind regards,
Nisha

Ben Vandervalk

unread,
Aug 2, 2016, 4:00:26 PM8/2/16
to Nisha pillai, ABySS, Esakimuthu Pillai Nisha PILLAI (IMCB)
Sorry for the slow response, Nisha.

The number of N's used to estimate the size gap does not matter to Sealer.  You could replace all the gaps with a single 'N' and the Sealer results would still be the same.  What matters most to Sealer is the true gap size.   Sealer does a bidirectional graph search starting from the flanks of the gap and tries to find a connecting path in the de Bruijn graph.   However, as the true gap size grows, this graph search becomes exponentially more expensive and eventually it will give up ("too many branches").   Thus Sealer is currently best at closing smaller gaps (<= 1000 bp).

I believe that in the Sealer paper, the gap-filling runs on human were done on an ABySS assembly of sequencing reads for individual NA19238 of the 1000 Genomes projects, and *not* on the reference genome.   That probably explains the discrepancy in the number of gaps that you see.

Hope that helps,

- Ben

--
You received this message because you are subscribed to the Google Groups "ABySS" group.
To unsubscribe from this group and stop receiving emails from it, send an email to abyss-users...@googlegroups.com.
For more options, visit https://groups.google.com/d/optout.

Reply all
Reply to author
Forward
0 new messages