Rail-RNA v0.1.6

Bug fixes:

• Compiling TSV with introns across samples (junctions.tsv.gz) no longer breaks when there are more than 9 samples.
• UnicodeEncodeErrors no longer appear in certain terminals.
• Versions of all dependencies (Bowtie 1, Bowtie 2, SAMTools, and PyPy) are now the same across all of Rail's modes; previously, versions of elastic mode dependencies were different from versions of local and parallel mode dependencies.
• Phred64 quality strings are now detected and converted to Sanger format during preprocessing; previously, they weren't interpreted properly.

Feature additions:

• A new cross-sample output counts.tsv.gz stores counts of primary alignments and uniquely aligning reads by chromosome.
• Median and mean coverages across samples normalized by --library-size (default: 40M) are now provided in bigWig format. By default, these average coverages are divided up by chromosome.
• A new option --skip-bad-records skips bad records when they are encountered during preprocessing.
• The installer now has backup URLs for dependencies to help ensure downloads succeed in the event of server failures.

Miscellaneous changes:

• Rail-RNA now installs itself in /usr/local/raildotbio or ~/raildotbio by default, and the directory structure of the installation is changed to facilitate uploading required software to S3 before running job flows in elastic mode.

Rail-RNA v0.1.5

Bug fix:

• Previously, reads with identical nucleotide sequences for which more than one exact end-to-end alignment to the genome was found were all assigned the same primary alignment. Such ties are now broken at random on a read-by-read basis.

Feature additions:

• A new --deliverables command-line option permits changing the final outputs of a job flow. The fewer outputs desired, the fewer steps Rail-RNA needs to execute.
• New cross-sample deliverables introns.tsv.gz, insertions.tsv.gz, and deletions.tsv.gz store the number of primary alignments overlapping/containing "significant" introns/indels. Here, "significant" means that the intron/indel is (by default) found in at least 5% of samples or overlapped by/contained in at least 5 primary alignments.
• Rail-RNA can now output only its transcript fragment ("isofrag") index for later reuse. Reusing an isofrag index is somewhat analogous to passing other aligners a transcript annotation: Rail-RNA skips several steps and does not search for introns, but rather aligns segments of reads to previously uncovered isofrags. A compressed (.tar.gz) isofrag index may be specified with the --isofrag-idx command-line parameter.
• Previously, only bigWigs with coverage by all primary alignments were output. Now, bigWigs with coverage by uniquely aligning reads (i.e., the scores of their primary alignments are not tied with the scores of any secondary alignments) are also output.
• A new -k command-line option permits specifying the maximum number of alignments output per read. The alignments are output in order of descending rank determined by Rail-RNA's objective. (See Eqs (1)-(3) of the preprint.)

Rail-RNA v0.1.0a

Bug fixes:

• AWS CLI local install no longer gives OSError
• IPython local install now works; hadn't previously been finding setup.py

Feature additions:

• Adds --no-direct-copy command-line parameter in elastic mode that copies intermediate data to HDFS before copying to S3 in a subsequent step

This version of Rail-RNA is used for the all-of-GEUVADIS run discussed in our preprint.

Run chmod +x install_rail-rna-0.1.0a, then ./install_rail-rna-0.1.0a to install with executable below.

Rail-RNA v0.1.0

First release.

Run chmod +x install_rail-rna-0.1.0, then ./install_rail-rna-0.1.0 to install with executable below.