SEQanswers

SEQanswers (http://seqanswers.com/forums/index.php)
-   Bioinformatics (http://seqanswers.com/forums/forumdisplay.php?f=18)
-   -   scaffolding without paired-end, mate pair (http://seqanswers.com/forums/showthread.php?t=21638)

yifangt 07-12-2012 01:40 PM

scaffolding without paired-end, mate pair
 
How do we scaffold contigs assembled from different phases of sequencing. Say I have:
1) BAC ends from Sanger sequencing, single end only (~1kb);
2) contigs (200bp ~ 500kb ) from 454 data paired end, the raw data has been deleted;
3) Contigs (100bp ~200kb) from Illumina PE, too. Assembled from CLCBio.
4) Some sequences from GenBank or my local deposite, maybe 10Mb;

Now I want to combined these 4 sets data together to build scaffolds. Is there package to do this type of job? Most packages handle short reads, not sure if there is one dealing with long reads of fasta format.

Read some posts in this forum, e.g. some said it impossible to do scaffolding without paired-end information. CAP3 seems to be the one to do the job, but was extremely slow for my first try. CLCbio can do the job, but not in command line format which does not help me for my case for ~7300 jobs (individual BAC clones).

Asking around for advice.

Thanks a lot!
YT

Wallysb01 07-12-2012 05:58 PM

You could pipe all the contigs and BACs through an assembly program again. You might want to chunk the contigs up into overlapping sudo-reads first though. Not having the raw data is pretty large disadvantage, especially if the contigs from the 454 and illumina data are really poor. This meta-assembly of different technologies is not an easy problem to overcome, and not having raw data limits your options.

yifangt 07-12-2012 11:35 PM

RE:
 
Is there a program pipeline to do this type of assembly beside CAP3?
Tried CAP3, but extremely slow. SSPACE/BAMBus/Phrap etc all need quality or PE information that I do not have.
Thanks!

Wallysb01 07-13-2012 09:22 AM

I've only used CAP3 for this meta-assembly stuff, and it is slow, but in my experience the week or two it took was worth it.

You could just make up quality values, say give them all a quality score of 25. You're already making pretty large modifications to the data by assembling it and reassembling it again anyway. Heck, many people doubt the validity of the estimated quality values from NGS machines anyway.

scalabrin 07-13-2012 09:52 AM

Assembly merge?
 
Quote:

Originally Posted by yifangt (Post 78754)
How do we scaffold contigs assembled from different phases of sequencing. Say I have:
1) BAC ends from Sanger sequencing, single end only (~1kb);
2) contigs (200bp ~ 500kb ) from 454 data paired end, the raw data has been deleted;
3) Contigs (100bp ~200kb) from Illumina PE, too. Assembled from CLCBio.
4) Some sequences from GenBank or my local deposite, maybe 10Mb;

YT

Dear YT,

I would try merging the last 3 datasets first. I would call it assembly merge or assembly reconciliation!

One tool to do it, under development in our lab is GAM (Genome Assembly Merger) that you can download from git:

git clone git://github.com/vice87/gam-ngs.git

What you need to do is to align a common dataset of reads (perhaps those you used for contigs3) to your different assemblies. At the moment you can work only pairwise. So I would suggest to merge set 2 with set 3 and then merge it with set 4.

The tool is still in its infancy so you can try it and I could provide you with some assistance, but not that much :-(

Finally, use BES to do real scaffolding (perhaps a module of Arachne or SSPACE can do that, but I am not sure they run independently from their whole pipeline).

Best regards,
Simone


All times are GMT -8. The time now is 10:07 AM.

Powered by vBulletin® Version 3.8.9
Copyright ©2000 - 2021, vBulletin Solutions, Inc.