Entering edit mode
12.4 years ago
Michael Kuhn
5.0k
I'm loosely following the progress of a nematode sequencing project where DNA was extracted from a few hundred individuals from the wild. Now, there are problems with the assembly of the reads, that may be caused by heterozygosity or polymorphisms of the population and potential contamination with another (sub)species.
- Are there any assemblers that can deal with this?
- Are there any tools that can look at the set of reads and "diagnose" the degree of heterozygosity?
- Would, in such a case, paired-end or mate-pair sequencing with longer insert sizes help?
(I'm imagining that there are stretches of sequencing where there are no big differences, but then there will be insertions/deletions that make it impossible to continue with the assembly.)
Pooling data is bad for de novo assembly. I doubt any assemblers can handle this well. Even a diploid ciona can cause great troubles when using sanger reads (the paper suggested by Casey Bergman), hundreds of worms sequenced with short reads will be much harder. If it were me, I would wait for cleaner data before doing any serious analysis.
Agreed, better to solve this at the source than try to deal with the mess from a poor sample/experimental design. For example, the D. simulans genome is also a mess because multiple strains were used in the assembly (informally referred to as "Franken-sim" http://bioinformatics.ufl.edu/mclab_public_data/README_updating_fusions.txt)
Pooling data is bad for de novo assembly. I doubt any assemblers can handle this well. Even a diploid ciona can cause great troubles when using sanger reads (the paper given by Casey Bergman), hundreds of worms sequenced with short reads will be much harder. If it were me, I would wait for cleaner data before doing any serious analysis.
Pooling data is bad for de novo assembly. I doubt any assemblers can handle this well. Even a diploid ciona can cause great troubles when using sanger reads (the paper given by Casey Bergman), hundreds of worms sequenced with short reads will be much harder. If it were me, I would wait for cleaner data before doing any serious analysis.
I wouldn't call it poor design, it's just an inherent problem with sequencing small uncultured animals. Still I agree that different wet-lab approaches need to be tried here.