Pvacseq

We added an optional downstream analysis tool to generate an annotated fasta file from a VCF with protein sequences of mutations and matching wildtypes. This tool can be run with the `pvacseq generate_protein_fasta` command.

This release fixes a couple of errors that were introduced in the previous version which would occur during the processing of certain inframe indels.

This version also fixes an error that would occur if the number of variants to process was a multiple of the chosen `--fasta-size`.

pVAC-Seq now supports local installs of IEDB MHC class I (http://tools.iedb.org/mhci/download/) and class II (http://tools.iedb.org/mhcii/download/) binding prediction tools. This feature can be used by passing the directory that contains the local installations to the `--iedb-install-directory` parameter.

This version adds a new column `Mutation Position` to the report output. This column denotes the 1-based start position of the mutation in the `MT Epitope Seq`. If the value is `0` the mutation start position is before the first position in the epitope.

pVAC-Seq now allows the user to specify the number of retries after a request to the IEDB RESTful interface fails. The number of retries can be set by using the `--iedb-retries` parameter. Previously this number was hard-coded to 3. More retries might be necessary in order to get a successful response for complex queries (e.g., large number of variants, long frameshift downstream sequences, choice of compute-intensive prediction algorithms). This parameter should be used in conjunction with `--fasta-size` and `--downstream-sequence-length` for the highest likelihood of success of finishing a pVAC-Seq run.

This release fixes an error that was introduced in the previous version which would occur when the user would try to rerun a process in the same output directory.

This version also fixes a bug with how to handle variants that are no-call or homozygous-reference. These variants will now be skipped.

This version fixes compatibility issues with the new IEDB Analysis Resource v2.15 that was released on October 3rd 2016. Since pVAC-Seq uses the IEDB RESTful web interface, the new IEDB Analysis Resource version is being used natively. As a result, pVAC-Seq output for the NetMHC prediction algorithm is now NetMHC version 4.0. Some other changes in the output format from IEDB led to compatibility issues that are fixed in this new pVAC-Seq release.

This release also fixes an error with bam-readcount input files that occurs when a variant does not have a bam readcount entry.

- Support for MHC Class II predictions
- Integration of NetChop cleavage site predictions and NetMHCstab stability predictions
- Support for bam-readcount and Cufflinks input files to provide coverage data

- Bugfix: The generation of the fasta file would fail for some insertions with a range position. This is now fixed.
- Bugfix: The generation of the fasta file would fail if the wildtype or downstream sequences were too long. The size limit for these fields has been increased to the user system's maximum supported size. This error might still occur if the sequences are longer than that.
- Bugfix: When rerunning a command an error would occur if the `tmp` subdirectory already exists in the output directory. This has now been fixed.

- Certain intermediate files are now written into a `tmp` directory underneath the main output directory. This `tmp` directory will be deleted at the end of a successful run unless the `--keep-tmp-files` flag is set.
Intermediate files will now not be reprocessed if they already exist in the output directory. This can be helpful if a run exits early, for example, when a 500 Error was returned by IEDB. In this case the user can now simply run the same `pvacseq run` command again and the run will pick up where it failed previously.
- We added a new option `--fasta-size` that the user can set to specify how many FASTA entries at a time will be included in a request to the IEDB RESTful API. The default is 200 but certain variants or prediction algorithms might warrant a smaller number of FASTA entries in order to avoid timeouts from IEDB.
- Bugfix: The parsing step would fail for frameshift mutations with a range position. This is now fixed.
- Bugfix: Previously a run might fail if certain intermediate files weren't created.
- Bugfix: Using `.` in the output directory name and the sample name would previously result in errors. This has now been fixed.
- Bugfix: Using a relative directory path for the output directory would previsouly result in an error. This is now fixed.