This is a recipe that provides a time- and memory- efficient way to loosely estimate the likely size of your assembled genome or metagenome from the raw reads alone. It does so by using digital normalization to assess the size of the coverage-saturated de Bruijn assembly graph given the reads …
read moreThere are comments.
This recipe provides a time-efficient way to determine whether you've saturated your sequencing depth, i.e. how much new information is likely to arrive with your next set of sequencing reads. It does so by using digital normalization to generate a "collector's curve" of information collection.
Uses for this recipe …
read moreThere are comments.
The below is a recipe for subsetting a high-coverage data set to a given average coverage. This differs from digital normalization because the relative abundances of reads should be maintained -- what changes is the average coverage across all the reads.
Uses for this recipe include subsampling reads from a super-high …
read moreThere are comments.
In recent days, we've gotten several requests, including two or three on the khmer mailing list, for ways to extract shotgun reads based on their coverage with respect to the reference. This is fairly easy if you have an assembled genome, but what if you want to avoid doing an …
read moreThere are comments.
We just released khmer v1.1, a minor version update from khmer v1.0.1 (minor version update: 220 commits, 370 files changed).
Cancel that -- _I_ just released khmer, because I'm the release manager for v1.1!
As part of an effort to find holes in our documentation, "surface" any …
read moreThere are comments.
As part of the 2-day Mozilla Science Labs hackathon in late July, the khmer project will be providing a "mentored open source contributathon" experience. This will provide an opportunity for people interested in trying out our instance of the "github flow" model, in which contributions are submitted for review using …
read moreThere are comments.
(or, What I Did For One Day Of My Summer Vacation.)
tl;dr? I played around with building a CountMin Sketch that is dynamic in size, based on a scalable Bloom Filter approach. I'm not sure it worked. Thoughts, suggestions, help?
In our research, we've made some hay …
There are comments.