I would like to announce that we have just released a new version of our Bisulfite-seq alignment and methylation calling tool Bismark. All associated files are available for free from http://www.bioinformatics.bbsrc.ac.uk/projects/.
As the most noticeable difference, Bismark does now further subdivide non-CpG context into CHG and CHH context, which will be especially interesting for researchers working on plant systems. The former characters 'C/c' in the methylation call has been replaced by:
CHG-context: X / x (methylated / unmethylated)
CHH-context: H / h (methylated / unmethylated)
In addition, I noticed that due to recent changes in the Bowtie source code, Bismark was producing lots of warnings 'best-first memory chunk exhaustion...') which was also mentioned here on SEQanswers. As suggested by Ben Langmead, the best way to counteract this problem is to increase the memory size for each bowtie thread, or mute bowtie. Thus, Bismark will now understand the additional option '--chunkmbs <int>' to change the memory from 64 (default)to any integer (I found that 256 or even 512 got rid of nearly all warnings). These errors were especially frequent in --best mode or for paired-end alignments. Bismark will now also understand the '--quiet' option to suppress memory chunk exhaustion (and other) warnings.
Some other minor fixes include:
- FastA files do no longer require the file ending ".fa".
- Fixed an issues so that Bismark will no longer tolerate chromosomes with
same name when reading the genome into memory.
- Fixed an issue with paired-end alignment reports.
- The methylation extractor will by default distinguish between cytosines in the three contexts CpG, CHG or CHH. If this is not needed, CHG and CHH context can be merged into 'non-CpG' context by specifying '--merge_non_CpG'.
- Due to the fundamental changes in v0.2.0 (CHG and CHH context methylation calls) the methylation extractor will now require that the Bismark mapping result file was generated with the same version of Bismark.
If you have any suggestion or comments I would like from you!
Best wishes,
Felix
As the most noticeable difference, Bismark does now further subdivide non-CpG context into CHG and CHH context, which will be especially interesting for researchers working on plant systems. The former characters 'C/c' in the methylation call has been replaced by:
CHG-context: X / x (methylated / unmethylated)
CHH-context: H / h (methylated / unmethylated)
In addition, I noticed that due to recent changes in the Bowtie source code, Bismark was producing lots of warnings 'best-first memory chunk exhaustion...') which was also mentioned here on SEQanswers. As suggested by Ben Langmead, the best way to counteract this problem is to increase the memory size for each bowtie thread, or mute bowtie. Thus, Bismark will now understand the additional option '--chunkmbs <int>' to change the memory from 64 (default)to any integer (I found that 256 or even 512 got rid of nearly all warnings). These errors were especially frequent in --best mode or for paired-end alignments. Bismark will now also understand the '--quiet' option to suppress memory chunk exhaustion (and other) warnings.
Some other minor fixes include:
- FastA files do no longer require the file ending ".fa".
- Fixed an issues so that Bismark will no longer tolerate chromosomes with
same name when reading the genome into memory.
- Fixed an issue with paired-end alignment reports.
- The methylation extractor will by default distinguish between cytosines in the three contexts CpG, CHG or CHH. If this is not needed, CHG and CHH context can be merged into 'non-CpG' context by specifying '--merge_non_CpG'.
- Due to the fundamental changes in v0.2.0 (CHG and CHH context methylation calls) the methylation extractor will now require that the Bismark mapping result file was generated with the same version of Bismark.
If you have any suggestion or comments I would like from you!
Best wishes,
Felix
Comment