Seqanswers Leaderboard Ad

Collapse

Announcement

Collapse
No announcement yet.
X
 
  • Filter
  • Time
  • Show
Clear All
new posts

  • Not getting satisfing results in SOAPde-novo

    Hi all,

    I'm using SAOP de-novo to assemble reads of a bacteria. We assess that the size of the genome should be around 100 - 200 kbp.
    We sequenced paired end reads 64bp X 2, and have a total of ~24M reads after filtering.
    For some reason SOAP de-novo doesn't give good results. I tried to use different K-mer values. I either get a cover of 1.8 Mbp with about 5,900 scaffolds and singletons (Kmer = 61), or a cover of 20,000 bp with 46 scaffolds and singletons.
    I also tried to put different map_len values (from 35 to 50) and it didn't change a lot.

    Shouldn't this be a simple task for SOAP de-novo? If the assessment of the genome size is correct, I have a coverage of 120X ...
    Why am I have problems?

    Thanks in advance,
    Rachelly.

  • #2
    Originally posted by Rachelly View Post
    Hi all,

    I'm using SAOP de-novo to assemble reads of a bacteria. We assess that the size of the genome should be around 100 - 200 kbp.
    We sequenced paired end reads 64bp X 2, and have a total of ~24M reads after filtering.
    For some reason SOAP de-novo doesn't give good results. I tried to use different K-mer values. I either get a cover of 1.8 Mbp with about 5,900 scaffolds and singletons (Kmer = 61), or a cover of 20,000 bp with 46 scaffolds and singletons.
    I also tried to put different map_len values (from 35 to 50) and it didn't change a lot.

    Shouldn't this be a simple task for SOAP de-novo? If the assessment of the genome size is correct, I have a coverage of 120X ...
    What filtering did you do - based on quality, adapters etc?

    SOAP, as with most short-read assemblers, ignores quality values (so low quality bases are treated the same as high quality ones) - for sure you want anything below Q10 to be gone, maybe up as far as Q20.

    In my experience, SOAP tends to work badly with k > 40-45% of the read length, and with only 64 base reads, you're certainly way too high with K=61.

    First, try to isolate the problem to either the assembly or scaffold stage - you'll need to get a decent assembly before optimizing the scaffolding.

    Comment


    • #3
      What filtering did you do - based on quality, adapters etc?
      I filtered the reads according to quality - I use only reads all their bases are above Q30.

      In my experience, SOAP tends to work badly with k > 40-45% of the read length, and with only 64 base reads, you're certainly way too high with K=61.
      Thanks for the tip, I will try to use lower K-mers.

      First, try to isolate the problem to either the assembly or scaffold stage - you'll need to get a decent assembly before optimizing the scaffolding.
      To do only the assembly part I should run only the pregraph and contig steps? How can assess what is a "decent assembly" ?

      Comment


      • #4
        Originally posted by Rachelly View Post
        I filtered the reads according to quality - I use only reads all their bases are above Q30.
        Strict - but i guess you have the data and application to allow that. Do you do anything about adapters?

        Originally posted by Rachelly View Post
        To do only the assembly part I should run only the pregraph and contig steps? How can assess what is a "decent assembly" ?
        Yes, just the pre-graph and contig steps.

        See what the resulting contigs look like in N50 / size terms. The N50 isn't likely to be very good (SOAP doesn't use paired end data during assembly), but the total size (ignoring really short contigs near the kmer length) should be approaching the same order of magnitude as the genome size (allowing for repeats which will be folded), given the high coverage you have.

        If you have any related genomes to compare against, that always helps too.

        Comment


        • #5
          Still doesn't work...

          Hi again,

          I still haven't managed to get a decent assembly.
          - I tried to use only reads from higher quality (Q37, Q38)
          - I filtered the reads from adapters (only about ~0.05% of the reads)
          - I filtered contaminates as much as I can think of
          - I played a lot with the K-mer value

          The bases covered is what I expect - around 100,000bp. But I always get around 200 scaffolds and a really low N50 - around 190bp..

          There's no reason this shouldn't work, it's only a small plasmid.
          Does anyone have an idea what else I could try?

          Thanks,
          Rachelly.

          Comment


          • #6
            Rachelly,

            How much of your data are you using for the assembly? Too much coverage can actually make a de novo assembly worse. I would limit my input to no more than 50X coverage, so for your 100,000bp plasmid this would be 5Mbp. This equates to ~39,000 of your 2x64bp reads. Use a script to randomly select this number of read pairs from your filtered, high quality reads and then try your assembly.

            Comment


            • #7
              Thanks for your reply kmcarr,

              Actually, I've tried that as well. I took only reads with quality of 38, that gave me a coverage of 55X, but the results didn't seem good. I got only 5k -6k bp covered..

              Rachelly.

              Comment


              • #8
                Hi !

                I am the author of Ray.

                Ray works very well on bacteria with paired end data.

                The maximum k-mer coverage allowed in Ray is 65535.



                Ray will happily assemble all of your reads. First, I suggest you try with a k-mer length of 21 -- this is the default in Ray.

                Ray is available here: http://denovoassembler.sf.net

                mpirun -np 8 Ray -k 21 -p forward.fastq reverse.fastq -o test-bacteria-ray


                Be sure that both forward.fastq and reverse.fastq contain the same number of sequences.

                After that, can you paste the content of these files:


                test-bacteria-ray.CoverageDistributionAnalysis.txt

                test-bacteria.ray-LibraryStatistics.txt


                -seb

                Comment


                • #9
                  So I assume since it is a bacteria that there aren't a lot of issues with repeats? How about high levels of structural heterozygosity in your pool of sequences?

                  What kind of insert sizes do you use?

                  One thing you could try doing is base level error correction as a pre-processing step with a program like Quake (do this with your reads prior to quality trimming, but after adapter removal):


                  You could also try a post processing scaffolding with a program like IMAGE:
                  Download IMAGE for free. IMAGE stands for Iterative Mapping and Assembly for Gap Elimination. It is a software designed to close gaps in any draft assembly using Illumina paired end reads.


                  which assembles the sets of reads that are unpaired at the ends of scaffolds and uses that to try and overlap new scaffolds. I don't know if I would trust it on a highly repetitive genome, but if yours doesn't have that issue then it seems like a good thing to try.

                  Also how do you strip off adapter sequences? There are various ways that adapters can appear. I have even seen them with the first base of the adapter missing and the read itself starting with the second base of the adapter (probably due to an adapter dimer with no genomic DNA present at all).


                  You could check out the Assemblathon manuscript for a genome assembly contest on a 100MB simulated genome with simulated reads. Many of the world's genome centers participated including BGI (but unfortunately not including the developers of Ray!). In the supplementary material the different genome assembly teams provide some information about the steps they use to do the assembly itself:
                  Last edited by jstjohn; 05-31-2011, 10:05 AM.

                  Comment

                  Latest Articles

                  Collapse

                  • seqadmin
                    Strategies for Sequencing Challenging Samples
                    by seqadmin


                    Despite advancements in sequencing platforms and related sample preparation technologies, certain sample types continue to present significant challenges that can compromise sequencing results. Pedro Echave, Senior Manager of the Global Business Segment at Revvity, explained that the success of a sequencing experiment ultimately depends on the amount and integrity of the nucleic acid template (RNA or DNA) obtained from a sample. “The better the quality of the nucleic acid isolated...
                    03-22-2024, 06:39 AM
                  • seqadmin
                    Techniques and Challenges in Conservation Genomics
                    by seqadmin



                    The field of conservation genomics centers on applying genomics technologies in support of conservation efforts and the preservation of biodiversity. This article features interviews with two researchers who showcase their innovative work and highlight the current state and future of conservation genomics.

                    Avian Conservation
                    Matthew DeSaix, a recent doctoral graduate from Kristen Ruegg’s lab at The University of Colorado, shared that most of his research...
                    03-08-2024, 10:41 AM

                  ad_right_rmr

                  Collapse

                  News

                  Collapse

                  Topics Statistics Last Post
                  Started by seqadmin, Yesterday, 06:37 PM
                  0 responses
                  8 views
                  0 likes
                  Last Post seqadmin  
                  Started by seqadmin, Yesterday, 06:07 PM
                  0 responses
                  8 views
                  0 likes
                  Last Post seqadmin  
                  Started by seqadmin, 03-22-2024, 10:03 AM
                  0 responses
                  49 views
                  0 likes
                  Last Post seqadmin  
                  Started by seqadmin, 03-21-2024, 07:32 AM
                  0 responses
                  67 views
                  0 likes
                  Last Post seqadmin  
                  Working...
                  X