This produces a refined VCF file with the most confident variant calls. One can filter based on many things, and the filtering we performed before was based on coverage, but I would now recommend that one uses things like mapping quality and genotype probability/quality scores to perform filtering, since this essentially encompasses the sequencing depth. Filter genotype calls by quality: Again, one could use SAMtools or GATK.I’m planning to write another post outlining the GATK approach in the context of reference-guided assembly. The former is more streamlined and quick to implement while the Best Practices approach of the latter is more refined (e.g., INDEL realignment), which might help in this type of application. Call any SNPs or INDELs: The two leaders here are SAMtools and GATK, which produce an output VCF file of any variants.I’m hoping to do a follow up post about this. This step can obviously vary a lot depending on the distance between the target and guide, and one has to balance high-quality mapping and sufficient coverage. Map quality-trimmed reads to a guide genome: I like BWA the most for this, but any mapper could work.Quality trim raw Illumina reads: Can use anything but I’m currently partial to Trimmomatic.When someone asked for guidance with their own data, I was quick to provide what I think is a more sound (and certainly more “open”) pipeline, which was facilitated by all I’ve learned since publishing the original paper. Export the consensus sequence for regions where reads mapped based on the level of read coverage.Map quality-trimmed reads to a guide genome. The basic pipeline from the before (i.e., the publication) was to use CLC Genomics Workbench to do the following: #Clc genomics workbench de novo assembly updateAdmittedly, at the time I was quite new to graduate school and to handling and analyzing whole-genome sequencing data, and being more seasoned now, I thought I would update my approach to one that makes more sense and uses some different tools. #Clc genomics workbench de novo assembly fullI’m also coming full circle and playing around with this approach again with some of the work I am currently doing. Not many have taken notice of this work (at least based on citations), though I recently had an inquiry from someone who was trying to implement it. Moreover, I think that even in cases of high sequencing coverage, where a de novo assembly is the norm, this type of approach could complement existing assembly methods to produce a better overall assembly. Despite this and other shortcomings, I hold out some hope that this type of assembly approach will become more common as more and more high-quality reference genomes from across the tree of life become available. Obviously, we were making the assumption that enough synteny exists to prevent our method from producing spurious assemblies, and this assumption may or may not hold up once some evidence starts emerging (it currently hasn’t across broader taxonomic scales). This was helped along by the fact that we were working with bird genomes, which are the best-case scenario for this type of approach in vertebrates. We found that with the 3-5x genome coverage we had, this approach produced a better genome assembly, based on several metrics, than simply performing de novo genome assembly, despite mapping to a guide genome that was quite distantly related in both cases. As the paper pointed out, this has been applied between strains of single species, but hasn’t really been tried across larger evolutionary distances. A (still in progress) best practices workflow for reference-guided genome assemblyĪ couple years ago I led a paper in PLoS ONE outlining a reference-guided genome assembly approach that, more-or-less, simply maps reads from the target species to a high-quality guide genome and exports a consensus, producing a genome sequence for the target species one is interested in.
0 Comments
Leave a Reply. |
AuthorWrite something about yourself. No need to be fancy, just an overview. ArchivesCategories |