Merging Velvet Assemblies

11-23-2010, 11:19 PM

Hi,

This is my first post, and I look forward to being part of the community.

I've been prepping, sequencing, and assembling pools of ~10 BAC clones using PE100 reads on an Illumina. The average clone size is ~150 kb, but they can range from 50-250 kb. During the preps, I pooled an equal weight of DNA for each BAC clone, so I expect different levels of coverage from each BAC. I know that Velvet produces optimal assemblies with a k-mer coverage of 20-30X, and k-mers of ~55 give me an average coverage of that level. However, large BACs will have <20X coverage and small BACs will have >30X with this k-mer. To deal with this, I've been running Velvet with a series of k-mers (31, 41, ..., 81), and my plan is to merge the contigs from the series of assemblies.

Initially, I just used Mummer to align the contigs produced from each assembly to the other assemblies, and I wrote a script to parse the Mummer output and discard contigs that are nested within larger contigs. This works OK, but I'm looking for something more sophisticated. Does anybody have suggestions of the best software for doing this merging. What I want to do is quite simple, but I'm just not sure of the best software to use.

Thanks,
Mike

siiner · 11-24-2010, 01:23 AM

I found some tools can do this job, such as CAP3, Phrap, CA, and MAIA.
But I didn't actually make any of them work well.
Hope you could try and show your results.

kbushley · 12-17-2010, 12:38 PM

Hello Mike,

I'm also trying to do this. I think CAP3 might be the best tool but am still exploring this...there is a guy in our department who's written a program using CAP3 to merge velvet and abyss assemblies. It might be of some use. Let me know if you've found any other solutions. You also are in the great state of Oregon...where are you located?

12-17-2010, 02:51 PM

Hi kbushley,

I've been using Minimus2 and am somewhat satisfied. I haven't tried CAP3 yet. I'm at the University of Oregon. Go Ducks!!!

Best,
Mike

kbushley · 12-17-2010, 03:44 PM

Thanks, I was reading up on that one today. Would you be willing to share your script that parses MUMmer output...that sound rather useful. Go Beaves -

.

12-17-2010, 03:56 PM

Sure. Get me your email address and I'll send them.

natstreet · 01-02-2011, 11:49 PM

I'd also be really interested to give the scripts a try, if possible as this is something I've been looking for a good solution to. Can I send my email address to get a copy?

gfmgfm · 01-03-2011, 11:15 AM

I also ran trans-abyss and velevt (with different k-mer), created a fasta file of all the assemblies I got (from the various velvet runs and from trans-abyss) and ran on that cap3.
What is the script for cap3 is doing?
Am I missing an important step?

rongmancai · 03-30-2011, 02:14 PM

Hi, mike

I sent you an email and discussed about the merge of contigs using Mummer. I am not sure you get it. No reply after I sent message. Hope to hear from you. Thanks.

Rongman

Seth · 06-06-2011, 10:13 PM

Hi Mike,

Have you tried Phrap? I'm assembling overlapping BACs recently and I've tried CAP3, Minimus2 and Phrap to remove the redundance of merged contigs, and Phrap works best.
But there is still redundance in the final assembly. I'd like to try Mummer next. Can you send me a copy of your script?

Thanks!
Seth

natstreet · 06-06-2011, 10:21 PM

Quote:

Originally Posted by Seth

....
But there is still redundance in the final assembly.....
Seth

Hi Seth

How do you assess redundancy and how do you determine when two contigs are redundant and should be merged rather than being too different to each other? I'm not sure what species you work with, but for our assemblies of highly heterozygous plants this is a huge issue. So far I've failed to find an option for achieving this that isn't horribly slow on large(ish) assemblies (400 Mbp +).

One thing I haven't yet tried is using PCAP as a replacement for CAP3. Has anyone tried it?

I would also be interested in a copy of the script if possible.

Seth · 06-07-2011, 01:43 AM

Quote:

Originally Posted by natstreet

Hi Seth

How do you assess redundancy and how do you determine when two contigs are redundant and should be merged rather than being too different to each other? I'm not sure what species you work with, but for our assemblies of highly heterozygous plants this is a huge issue. So far I've failed to find an option for achieving this that isn't horribly slow on large(ish) assemblies (400 Mbp +).

Hi,
I used the total base count of final assembly to assess the redundancy. And the maximum length of target region can be estimated from the insert length of the BAC and BACs' count. I'm not familiar with the algorithms adopted in those softwares but I think the main idea is to identify overlapping contigs and join them together.

Have you tried Hapsembler? Designed for assembling highly heterozygous genomes, but also slow.

natstreet · 06-07-2011, 07:19 AM

Hi Seth

Thanks for the pointer to hapsembler, I hadn't come across it before. I'll test it out asap.

priya.s · 07-05-2011, 10:39 AM

Hi,

I am assembling Hepatitis C virus hypervariable regions E1 and E2, which have lots of SNPs. I am using hapsembler but it is very slow. How is your experience with hapsembler?

ZhigangLi · 09-26-2011, 06:38 PM

I have a similar problem. I want to combine contigs/scaffolds assembled with different dataset, e.g. sanger,454 and solexa. I wanted to combine them based on Mummer alignments. However, it's so hand for me. The organism is 40M and the largest scaffold is 2M. Can I use these software to finish my job?

rskr · 09-27-2011, 08:04 AM

Quote:

Originally Posted by ZhigangLi

I have a similar problem. I want to combine contigs/scaffolds assembled with different dataset, e.g. sanger,454 and solexa. I wanted to combine them based on Mummer alignments. However, it's so hand for me. The organism is 40M and the largest scaffold is 2M. Can I use these software to finish my job?

Cap3 and Mummer have issues in scaling, besides cap3 is for ESTs anyway.

anli · 10-03-2011, 04:12 AM

Hi.

I also have come across this issue. I have illumina PE data from an archaea and I now have two datasets were one is done with 100bp read-length and the other one with 150bp.

Doing an assembly on a merged dataset doesn't seem as a good approach, since you can't set multiple k-mer lengths in velvet.

boetsie · 10-03-2011, 02:53 PM

What do you guys want to do exactly? Do you want to make a consensus of the assemblies, or do you want to extend one of the assemblies by other assemblies?

You should be aware that you can merge repeated regions if they are at the boundaries of the contigs, and thus concatenate distant regions because of the repeat.

Anyway, what you can do is break the assemblies into smaller pieces and do a new denovo. I have a perl script which breaks all assemblies in user-defined k-mers and tries to do a new de novo assembly based on the users 'coverage'. Say you have four assemblies with different k-mers, and you only want to extend a contig by a k-mer if it is supported by e.g. three assemblies.

If you would like to have it, please contact me at marten.boetzer@baseclear.com

Regards,
Boetsie

edge · 01-07-2012, 12:10 AM

Hi Seth,

I got few question regarding CAP3 might need your advice.
I'm currently facing the following problem when trying to form a single set of non-redundant unigenes by CAP3

I have total of 8 *.fasta right now (RNA-seq scaffold sequence that extracted from same tissue but treated the sample with different condition for sequencing).
I would like to use CAP3 to assemble all the unigenes from different samples (but same tissue just treated the sample with different condition for sequencing) to form a single set of non-redundant unigenes.

Can I know what is the proper command I should apply when running CAP3 in order to form a single set of non-redundant unigenes of my RNA-seq data?
All my 8 sample scaffold in fasta format which is assembled by third party assembler program, Illumina pair-end read, 2X50bp, insert size 200.

This is all the info I have right now.
Many thanks for any advice.

milo0615 · 02-26-2015, 12:15 PM

Quote:

Originally Posted by kbushley

Hello Mike,

I'm also trying to do this. I think CAP3 might be the best tool but am still exploring this...there is a guy in our department who's written a program using CAP3 to merge velvet and abyss assemblies. It might be of some use. Let me know if you've found any other solutions. You also are in the great state of Oregon...where are you located?

Hi kbushley,

I am also trying to merge six different k-mer assemblies from Abyss. Were you able to merge yours? Can you share the program the guy from your department wrote? Please let me know. I would really appreciate your help.

Thank you,

-Milo

Similar Threads
Thread	Thread Starter	Forum	Replies	Last Post
Merging Assemblies with Minimus2	SLB	Bioinformatics	2	11-02-2014 01:50 PM
merging scaffolds from several SOAPdenovo assemblies into a single consensus assembly	jstjohn	Bioinformatics	10	08-13-2014 05:39 AM
Comparing and merging genome assemblies	megh	Bioinformatics	5	07-23-2014 04:58 AM
Merging genomic Assemblies	rahularjun86	De novo discovery	0	02-03-2012 03:08 AM
Merging Velvet Assemblies	millermr	General	0	11-23-2010 09:59 PM

11-23-2010, 11:19 PM	#1
millermr Guest Posts: n/a	Merging Velvet Assemblies Hi, This is my first post, and I look forward to being part of the community. I've been prepping, sequencing, and assembling pools of ~10 BAC clones using PE100 reads on an Illumina. The average clone size is ~150 kb, but they can range from 50-250 kb. During the preps, I pooled an equal weight of DNA for each BAC clone, so I expect different levels of coverage from each BAC. I know that Velvet produces optimal assemblies with a k-mer coverage of 20-30X, and k-mers of ~55 give me an average coverage of that level. However, large BACs will have <20X coverage and small BACs will have >30X with this k-mer. To deal with this, I've been running Velvet with a series of k-mers (31, 41, ..., 81), and my plan is to merge the contigs from the series of assemblies. Initially, I just used Mummer to align the contigs produced from each assembly to the other assemblies, and I wrote a script to parse the Mummer output and discard contigs that are nested within larger contigs. This works OK, but I'm looking for something more sophisticated. Does anybody have suggestions of the best software for doing this merging. What I want to do is quite simple, but I'm just not sure of the best software to use. Thanks, Mike

12-17-2010, 03:44 PM	#5
kbushley Member Location: Oregon Join Date: Jan 2010 Posts: 22	Thanks, I was reading up on that one today. Would you be willing to share your script that parses MUMmer output...that sound rather useful. Go Beaves -.

03-30-2011, 02:14 PM	#9
rongmancai Junior Member Location: Virginia Join Date: Mar 2011 Posts: 1	merge contigs Hi, mike I sent you an email and discussed about the merge of contigs using Mummer. I am not sure you get it. No reply after I sent message. Hope to hear from you. Thanks. Rongman

07-05-2011, 10:39 AM	#14
priya.s Junior Member Location: United States Join Date: May 2011 Posts: 1	Hapsembler Hi, I am assembling Hepatitis C virus hypervariable regions E1 and E2, which have lots of SNPs. I am using hapsembler but it is very slow. How is your experience with hapsembler?

11-24-2010, 01:23 AM	#2
siiner Junior Member Location: Shenzhen Join Date: Mar 2010 Posts: 2	I found some tools can do this job, such as CAP3, Phrap, CA, and MAIA. But I didn't actually make any of them work well. Hope you could try and show your results.

12-17-2010, 12:38 PM	#3
kbushley Member Location: Oregon Join Date: Jan 2010 Posts: 22	Hello Mike, I'm also trying to do this. I think CAP3 might be the best tool but am still exploring this...there is a guy in our department who's written a program using CAP3 to merge velvet and abyss assemblies. It might be of some use. Let me know if you've found any other solutions. You also are in the great state of Oregon...where are you located?

12-17-2010, 02:51 PM	#4
millermr Guest Posts: n/a	Hi kbushley, I've been using Minimus2 and am somewhat satisfied. I haven't tried CAP3 yet. I'm at the University of Oregon. Go Ducks!!! Best, Mike

12-17-2010, 03:56 PM	#6
millermr Guest Posts: n/a	Sure. Get me your email address and I'll send them.

01-02-2011, 11:49 PM	#7
natstreet Member Location: Sweden Join Date: Nov 2009 Posts: 83	I'd also be really interested to give the scripts a try, if possible as this is something I've been looking for a good solution to. Can I send my email address to get a copy?

01-03-2011, 11:15 AM	#8
gfmgfm Member Location: il Join Date: Jun 2010 Posts: 64	I also ran trans-abyss and velevt (with different k-mer), created a fasta file of all the assemblies I got (from the various velvet runs and from trans-abyss) and ran on that cap3. What is the script for cap3 is doing? Am I missing an important step?

06-06-2011, 10:13 PM	#10
Seth Junior Member Location: KM Join Date: Jul 2010 Posts: 4	Hi Mike, Have you tried Phrap? I'm assembling overlapping BACs recently and I've tried CAP3, Minimus2 and Phrap to remove the redundance of merged contigs, and Phrap works best. But there is still redundance in the final assembly. I'd like to try Mummer next. Can you send me a copy of your script? Thanks! Seth

06-07-2011, 07:19 AM	#13
natstreet Member Location: Sweden Join Date: Nov 2009 Posts: 83	Hi Seth Thanks for the pointer to hapsembler, I hadn't come across it before. I'll test it out asap.

09-26-2011, 06:38 PM	#15
ZhigangLi Member Location: Beijing, China Join Date: Nov 2010 Posts: 11	I have a similar problem. I want to combine contigs/scaffolds assembled with different dataset, e.g. sanger,454 and solexa. I wanted to combine them based on Mummer alignments. However, it's so hand for me. The organism is 40M and the largest scaffold is 2M. Can I use these software to finish my job? Last edited by ZhigangLi; 09-26-2011 at 07:09 PM.

10-03-2011, 04:12 AM	#17
anli Junior Member Location: Sweden Join Date: Apr 2011 Posts: 3	Hi. I also have come across this issue. I have illumina PE data from an archaea and I now have two datasets were one is done with 100bp read-length and the other one with 150bp. Doing an assembly on a merged dataset doesn't seem as a good approach, since you can't set multiple k-mer lengths in velvet.

10-03-2011, 02:53 PM	#18
boetsie Senior Member Location: NL, Leiden Join Date: Feb 2010 Posts: 245	What do you guys want to do exactly? Do you want to make a consensus of the assemblies, or do you want to extend one of the assemblies by other assemblies? You should be aware that you can merge repeated regions if they are at the boundaries of the contigs, and thus concatenate distant regions because of the repeat. Anyway, what you can do is break the assemblies into smaller pieces and do a new denovo. I have a perl script which breaks all assemblies in user-defined k-mers and tries to do a new de novo assembly based on the users 'coverage'. Say you have four assemblies with different k-mers, and you only want to extend a contig by a k-mer if it is supported by e.g. three assemblies. If you would like to have it, please contact me at marten.boetzer@baseclear.com Regards, Boetsie

01-07-2012, 12:10 AM	#19
edge Senior Member Location: China Join Date: Sep 2009 Posts: 199	Hi Seth, I got few question regarding CAP3 might need your advice. I'm currently facing the following problem when trying to form a single set of non-redundant unigenes by CAP3 I have total of 8 *.fasta right now (RNA-seq scaffold sequence that extracted from same tissue but treated the sample with different condition for sequencing). I would like to use CAP3 to assemble all the unigenes from different samples (but same tissue just treated the sample with different condition for sequencing) to form a single set of non-redundant unigenes. Can I know what is the proper command I should apply when running CAP3 in order to form a single set of non-redundant unigenes of my RNA-seq data? All my 8 sample scaffold in fasta format which is assembled by third party assembler program, Illumina pair-end read, 2X50bp, insert size 200. This is all the info I have right now. Many thanks for any advice.