methpipe-3.4.0: Tuco

Bug fixes

• Fixed bug in bsrate causing abort while processing reads that hang over the end of a chromosome (and added warning message).
• The mutation tracking in methcounts introduced a bug in merge-methcounts when parsing mutated contexts. merge-methcounts now only counts non-mutated site information and doesn't break while parsing mutated contexts.
• duplicate-remover statistics tracking now functions properly: previously the good_bases_out value was not correct when using the sequence info option.
• allelicmeth no longer throws a "could not convert" exception when converting the index of the last CpG back to genomic coordinates.
• hmr and pmd no longer skip the last domain of each chromosome
• merge-bsrate output now includes previously omitted headers, consistent with bsrate output
• Fixed hypermr numerical issues by renormalizing learned posterior and transition probabilities at each iteration of the Baum Welch training.
• to-mr now ignores discordant pairs

Enhancements

• WALT, our new space-seeded wildcard bisulfite read mapper, has been integrated into our manual as the de facto mapper in methpipe. It can be cloned from https://github.com/smithlabcode/walt
• radmeth, a recently developed tool for multi-factor, multi-replicate differential methylation analysis, has been integrated into methpipe. A detailed description of the functionality is available in the manual.
• bigwig_to_methcounts.py introduced as a tool to convert tracks downloaded from MethBase on the UCSC genome browser to methcounts format. A description is in the manual.
• methcounts memory usage halved and now prints estimate of memory usage for each chromosome
• methcounts now prints every cytosine in the reference by default, even if an entire chromosome is not covered, to maintain line number consistency across samples
• merge-methcounts now provides an option to output the merged methylomes in a table format for easy piped downstream analysis. It now prints a union of CpG sites from its input files, even though the number of CpG sites for each file is different.
• roimethstat memory usage reduced drastically by loading only CpGs within target regions into memory.
• Added option to hmr to specify random number generator seed, allowing user to exactly reproduce results if necessary.
• hmr now reports the "effective genome proportion," the percentage of the genome not in deserts, in the verbose output.
• hmr no longer has nondeterministic behavior.
• levels output format is slightly changed. Now it is technically in YAML format.

Organizational changes

• Substantial changes to the manual to introduce WALT as our new standard read mapper
• Removed experimental directory build from makefile to prevent user use of programs that are not rigorously tested or production-ready
• Some outdated and MethBase related stuff were removed.