merging scaffolds from several SOAPdenovo assemblies into a single consensus assembly

[jstjohn]

Does anyone have experience with merging scaffolds from a few assemblies into a single consensus assembly? When I run SOAPdenovo with different k cutoffs I get different resulting scaffold characteristics. Higher k values tend to give me a few longer scaffolds with better n50, but overall a lower mean scaffold size and more shorter scaffolds than with smaller values of k. It seems like a good idea to merge assemblies generated with a few of these kmer values into a single consensus assembly, right?

I have looked at a few tools to do this. One of them is called Reconciliator http://www.genome.umd.edu/reconcilia...structions.htm but it would require me to convert a few SOAPdenovo assembly output files into "Sanger/WashU" format which looks like it could involve quite a bit of work. I am not sure the amount of work required to generate all of these required input files would be worth it?

I also found minimus2 which looks like it is primarily designed to merge two assemblies at a time, and I have 4 I would like to merge. I could do pairwise merging with the previous consensus, but that seems like it could lead to problems... I am also finding some complaints on forums about how the program deals with N's.

The last program I found is called MAIA, but it is distributed as a Matlab package with dependencies on a few matlab distributed toolkits (yuck) and it looks like it also requires a "closely related reference genome" which I definitely do not have.

Thanks for any suggestions or experiences with this.

-John

[natstreet]

Another option is the recently-released Zorro (which is based on minimus2 but makes using NGS data friendlier). However, it is also for pairwise merging.

[boetsie]

maybe have a look at older threads;

http://seqanswers.com/forums/showthread.php?t=8045&highlight=merge

http://seqanswers.com/forums/showthread.php?t=9908&highlight=merge

and zorro looks also as an interesting tool!
Boetsie

[oben]

Hi~

Maybe this TGICL is helpful

http://compbio.dfci.harvard.edu/tgi/software/

Oben

[boetsie]

Another tool is GAM;

http://www.dmi.units.it/~casagran/research/software/software.html

haven't tried it yet though

[scalabrin]

Another tool is GAM;

http://www.dmi.units.it/~casagran/research/software/software.html

haven't tried it yet though

Hi,
GAM was sanger based and currently supports assembly from Arachne and PCAP only. We are working on a NGS version that supports bam files.

[luc]

GAM-NGS was published a few weeks ago:

https://github.com/vice87/gam-ngs

http://www.biomedcentral.com/1471-2105/14/S7/S6

It looks very promising; I needed the help from a very good admin to get it installed. For my large datasets however it stopped repeatedly about 20 minutes into one of the last steps.

[scalabrin]

GAM-NGS was published a few weeks ago:

https://github.com/vice87/gam-ngs

http://www.biomedcentral.com/1471-2105/14/S7/S6

It looks very promising; I needed the help from a very good admin to get it installed. For my large datasets however it stopped repeatedly about 20 minutes into one of the last steps.

Please, contact the contact author, he will surely help you!

About installation, strange, I was easily able to install it myself, even on my home without asking for any system path.

[mberacochea]

Hi,

I have a related question. I have a draft genome (sequenced with PGM) and I want to get the most of the data. Merging different assemblies (different software) is useful?

Thanks!

[scalabrin]

Hi,

I have a related question. I have a draft genome (sequenced with PGM) and I want to get the most of the data. Merging different assemblies (different software) is useful?

Thanks!

It depends on the assemblies you want to merge. If they are very similar to each other then there is no meaning to merge them. For example, if you run different assemblies with ABySS at different k-mer length, short kmer assemblies are usually subsets of longer kmers assemblies. While, if you are comparing an assembly done with AllPaths-LG and one with ABySS you might get different results and merging them would be very usefull.

[mberacochea]

I thought so, I haven't assembled the genome with other tools. Will do to check.

Thanks!