3.1.1

This is a bugfix release. It fixes the following problem(s):

• The --exclude-NAs flag was not being passed along correctly to the main pipeline and didn’t have any effect on downstream filtering.
• The aggregate report creation step had some inefficiencies which caused its runtime to be much longer than necessary.
• An unneeded import statement for the PyVCF package had the potential to cause errors while running the pVACseq pipeline.

3.1.0

This release adds the following new features:

• When running the pipelines with the --netmhc-stab flag enabled, the NetMHCstab allele column now also reports the distance of the NetMHCstabpan allele when that allele is not identical to the HLA Allele of the original result.
• When running the pipelines with a set of individual class II alleles, the pipeline now also auto-generates valid combinations of these alleles so that users no longer need to explicitly provide these combinations.

This release also fixes the following problem(s):

• Some class I alleles are not supported by NetMHCstabpan and will lead to an error when attempt to make predictions with them. This release will skip such alleles for the stability prediction step.
• For very large result sets the filtering steps would stall or be killed because the steps would run out of memory. This release fixes this issue.
• This release adds better handling of timeout errors while running NetMHCstabpan and/or NetChop.

3.0.4

This is a bugfix release. It fixes the following problem(s):

• This fixes an issue introduced in the previous release where VCF entries with no VAF value would result in an error.
• This release adds a new constraint to the vaf cutoff command line arguments to ensure that they are a fraction between 0 and 1.
• This release also fixes an issue where the wrong binding filter class was being used when running pVACfuse with allele-specific binding cutoffs.

3.0.3

This is a bugfix release. It fixes the following problem(s):

• When using the --additional-report-columns parameter in pVACview/pVACfuse, no contents were previously written to the additional columns.
• MHCflurry may return no value for the percentile binding score. This would previously result in an error which has been fixed in this release.
• Variants in a VCF may contain an empty AF field, which was previously not being handled correctly in all cases, resulting in an error.

This release also includes some minor improvements:

• When running the pVACseq pipeline there would be a lot of warning messages about missing position column parameters. These would mostly be noise because the underlying consequence type wasn’t supported by pVACseq to begin with. This release removes these warning messages for mutations with consequences that are not supported by pVACseq.

3.0.2

This is a bugfix release. It fixes the following problem(s):

• Newer versions of Ensembl annotations added genes without a HUGO gene name. Variants on those genes cause downstream errors in the aggregate report creation when running the pVACseq pipeline. This release fixes this error.

3.0.1

This is a bugfix release. It fixes the following problem(s):

• There was small issue in pVACview where picking a different number of variants in the main table would cause changes to variant evaluations to not be recorded correctly.
• This release also fixes a problem where loading a new set of files into pVACview would not reset previous evaluation summary counts.
• This release fixes an error in pVACtools that would occur when a chromosomal region in the somatic VCF is not present in the proximal VCF.
• This release adds further error checking for malformed CSQ header descriptions.
• This release updates various code portions to remove deprecation warnings and add support for newer versions of certain dependencies (e.g. pandas).
• This release fixes an bug with the standalone pvacseq calculate_reference_proteome_similarity command where picking a custom --match-length would throw an error.
• Some users were reporting errors during the tmp file cleanup stages of running the various pipelines. This releases fixes that issue.

3.0.0

This version adds the following features, outlined below. Please note that pVACtools 3.0 is not backwards-compatible and certain changes will break old workflows.

Breaking Changes

• The pVACapi and pVACviz tools have been removed. They have been replaced by the pVACview tool.
• The package namespace has been updated. The files will now be installed underneath a pvactools directory in your python package installation path.
• The aggregated report format has been updated. The headers have been updated for clarity. An additional column Allele Expr has been added, representing RNA expression * RNA VAF. For more information see all_epitopes.aggregated.tsv Report Columns
• pVACfuse no longer supports inputs from Integrate NEO. Only AGFusion inputs will be supported going forward.
• The format of the pVACfuse all_epitopes and filtered reports has been updated to remove columns that aren't applicable for the tool. Please see the pVACfuse output file documentation for more information.

New Features

• This release adds a new tool, pVACview. pVACview is an R Shiny application that allows for that visualization of the pVACseq aggregated report file to review, explore, and prioritize the different neoantigen candidates predicted by pVACseq.
• The 3.0 release adds several improvements to the reference proteome similarity step:
  • Users can now run the reference proteome similarity step with a standalone Protein BLAST installation. To use a standalone BLASTp installation, provide the installation path using the --blastp-path parameter. The supported Protein BLAST databases are refseq_select_prot and refseq_protein. Installation instructions for BLAST can be found here.
  • When running the reference proteome similarity step using the NCBI Protein BLAST API, users can now pick between the refseq_select_prot and refseq_protein databases.
  • Parallelization has been added to the reference proteome similarity step. When running this step as part of the pVACseq, pVACfuse, or pVACbind pipelines, the existing --t parameter will also be used to set the number of parallel threads in this step.
• This release adds standalone commands to run stability predictions, cleavage site predictions, and the reference proteome similarity step on the output of the pVACseq, pVACfuse, and pVACbind pipelines.

Minor Updates

• Previously, when running NetChop for cleavage site predictions, predictions were made for each epitope individually. However, these predictions will differ if additional flanking amino acids are provided and will be stable with 9 or more flanking amino acids. We updated this step to make predictions with 9 flanking amino acids around each epitope to generate stable predictions.
• This release adds a --species option to the valid_alleles commands to filter alleles on a species of interest.
• This release adds a --pass-only flag to the pvacseq generate_protein_fasta command to only process VCF entries that do not have a FILTER set.
• This release adds a new parameter --tumor-purity. This parameter indicates the fraction of tumor cells in the tumor sample and is used during aggregate report creation for a simple estimation whether variants are subclonal or clonal based on VAF.

2.0.7

This is a bugfix release. It fixes the following problem(s):

• This releases fixes an edge case that would result in an error when the proximal variant VCF didn’t contain a region from the somatic VCF.

2.0.6

This is a bugfix release. It fixes the following problem(s):

• When running pVAcseq with a proximal variants VCF, proximal DNPs affecting multiple amino acids were not handled correctly and would result in an error. This issue has now been fixed.

2.0.5

This is a bugfix release. It fixes the following problem(s):

• Some users have reported “Cannot open file” errors when running NetMHCstabpan. This release adds a retry when this error in encountered.
• This release adds stricter checking to pVACbind for unsupported amino acids. Sequences containing an unsupported amino acid will be skipped. The following amino acids are supported: A, R, N, D, C, E, Q, G, H, I, L, K, M, F, P, S, T, W, Y, V.
• Some VEP predictions for supported variant types might not contain any protein position information, rendering pVACseq unable to parse such annotations. Annotations without protein position information will now be skipped.