Change the default k to 64, add 32 and 96 #2

Merged
merged 1 commit into from Oct 3, 2014

Conversation

Projects
None yet
2 participants
@sjackman
Contributor

sjackman commented Oct 3, 2014

Hi, Michael. The default value of k of 25 is much too low for most data sets. This patch increases the default value of k to 64, and adds 32 and 96 as low- and high-k alternatives. We're looking into integrating KmerGenie or similar into the ABySS docker file. Have you considered integrating a k-choosing tool into Nucleotid.es upstream of all dBG assemblers?

This patch replaces #1—though I prefer #1 if you'll consider it, as it's more similar to our typical workflow with ABySS.

michaelbarton added a commit that referenced this pull request Oct 3, 2014

Merge pull request #2 from sjackman/k64
Change the default k to 64, add 32 and 96

@michaelbarton michaelbarton merged commit 27dbf4c into bioboxes:master Oct 3, 2014

@michaelbarton

This comment has been minimized.

Show comment
Hide comment
@michaelbarton

michaelbarton Oct 3, 2014

Contributor

Thank you for the pull request Shaun. I have merged this.

Thank you for adding a default to the Profile, I prefer that each image has
a default so that a non-technical can use an image and get a reasonable
assembly back. My objection to #1 was that a user would have to run each of
the kmer sizes and then determine which provided the best assembly. I think
if you wanted to search through kmer sizes then this should be done behind
the scenes inside the container and then return the best contigs to the
user.

I also agree that KmerGenie could be another good alternative. You could
run this on the fastq and use the estimated kmer value for running ABySS. I
don't plan to provide any additional information for containers, such kmer
selection, for the same reason above. I prefer each that image be simple
for a non-technical user so that they can just run it with their fastq and
get an assembly without additional work.

Contributor

michaelbarton commented Oct 3, 2014

Thank you for the pull request Shaun. I have merged this.

Thank you for adding a default to the Profile, I prefer that each image has
a default so that a non-technical can use an image and get a reasonable
assembly back. My objection to #1 was that a user would have to run each of
the kmer sizes and then determine which provided the best assembly. I think
if you wanted to search through kmer sizes then this should be done behind
the scenes inside the container and then return the best contigs to the
user.

I also agree that KmerGenie could be another good alternative. You could
run this on the fastq and use the estimated kmer value for running ABySS. I
don't plan to provide any additional information for containers, such kmer
selection, for the same reason above. I prefer each that image be simple
for a non-technical user so that they can just run it with their fastq and
get an assembly without additional work.

@sjackman

This comment has been minimized.

Show comment
Hide comment
@sjackman

sjackman Oct 3, 2014

Contributor

Thanks, Michael. Feel free to close #1.

Contributor

sjackman commented Oct 3, 2014

Thanks, Michael. Feel free to close #1.

@sjackman sjackman deleted the sjackman:k64 branch Jun 13, 2015

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment