version-1.09

This version of corset redistributes reads when the -l flag is used rather than throwing them away. This option can help on large assemblies / datasets when computational resources are an issue.

version-1.08

This version:

• Includes a major update to improve computational efficiency (less memory and faster run time). This should make corset much easier to run on large assemblies.
• It introduces the flag "-I" to be used when the end goal is differential isoform usage analysis.
• If fixes a minor bug related to the p-value threshold for splitting clusters.

Since this version introduces some major changes to the hierarchical clustering data structure, there's a chance it may have also introduced unforeseen bugs. Please report any issues on our google group or through gitHub's "issues" tab.

version-1.07

This version offers compatibility with the Salmon aligner (making alignment and clustering much faster). See the wiki documentation for instructions.

version-1.06

This version introduces filtering options when corset format reads (.corset-reads) are used as input.

Specifically you can filter out any read which aligns to more than N contigs (through the option "-x "). For example, if a read aligns to more than 500 contigs it might indicate it is low in complexity and aligning to repeat sequence. We also introduce an option to filter on links between contigs. Where a "link" is created if a read maps to two more contigs. Links can be filtered out if they do not have enough supporting reads. For example, if only one reads maps to both contigA and contigB, it's possible these contigs do not belong in the same gene. You can filter using the option "-l ". A reasonable number for would be below 10. We now also print out the number of reads which pass/fail these filters.

These filtering options are likely to speed up the clustering in many cases.

version-1.05

This release introduces a feature to write out intermediate read summary files - a compact representation of the information in the bam files. This allows the bam reading phase of corset to be run in parallel with one job per sample. The compact read summary files can be read back into corset much quicker than bam files. This version also gives some improvements on the speed of reading the bam files more generally.

version-1.04

from 21st December 2014 (bug fix to allow compilation against samtools 1.1 and lower, otherwise identical to corset 1.03)

version-1.03

from 10th September 2014 (bug fix to allow compilation against samtools 1.0 and lower)