pVACtools is a cancer immunotherapy tools suite consisting of the following tools:
A cancer immunotherapy pipeline for identifying and prioritizing neoantigens from a VCF file.
A cancer immunotherapy pipeline for identifying and prioritizing neoantigens from a FASTA file.
A tool for detecting neoantigens resulting from gene fusions.
A tool designed to aid specifically in the construction of DNA-based cancer vaccines.
A browser-based user interface that assists users in launching, managing, reviewing, and visualizing the results of pVACtools processes.
New in release 2.0.4¶
This is a bugfix release. It fixes the following problem(s):
Failed calls to the NetChop and NetMHCstab API were not being caught correctly because failures would still result in a 200 return code. This would ultimately result in empty filtered report files. This release adds more error checking around the returned content from these APIs and will fail if the content is not formatted as expected.
This release adds handling of some more VCF edge cases that were previously unsupported. Variant transcripts that are annotated with * in the wildtype protein sequence or that have a stop_retained_variant consequence are now skipped. In addition, some variants may encode their postion as
-/1234, which was previsouly not supported but has now been added.
When running pVACseq, pVACbind, or pVACfuse with the
--run-reference-proteome-similarityoption enable this step would create a reference matches file but the pipeline previously failed to copy this file into the output directory. This release fixes that issue.
keras is now pinned to version 2.4.3 since newer versions might not be compatible with the pinned tensorflow version.
New in version 2.0¶
This version adds the following features, outlined below. Please note that pVACtools 2.0 is not backwards-compatible and certain changes will break old workflows.
pVACtools now supports variable epitope lengths for class II prediction algorithms. The previous option
-e) no longer exists. It has been replaced with
-e2) for class I and class II epitope lengths, respectively. The defaults are
[8, 9, 10, 11]and
[12, 13, 14, 15, 16, 17, 18], respectively.
--peptide-sequence-lengthoption has been removed. The peptide sequence length is now determined by the epitope length(s) to determine the flanking sequence length before and after the mutation.
pVACtools no longer depends on conda. pVACtools remains compatible with Python 3.5 and above but users may chose any environment manager to set up an appropriate Python environment.
When using standalone IEDB, pVACtools is now only compatible with IEDB 3.1 and above. Please see Installation for instructions on installing the latest IEDB version.
pVACseq is no longer dependent on annotations with the VEP Downstream plugin. This dependency has been replaced with the VEP Frameshift plugin. This requires changes to your existing VEP installation in order to install the Frameshift plugin. Existing VCFs that were previously annotated to work with pVACtools 1.5 and below will no longer work with version 2.0 and above and will need to be reannotated. Please see our documentation on Annotating your VCF with VEP for more information.
The filtered.condensed.tsv report has been removed and replaced with the all_epitopes.aggregated.tsv report. We believe that this new report will provide a more useful summary of your results. Please see the Output Files sections of each tool for more information on this new report.
pVACtools now provides binding affinity percentile rank information, in addition to the raw ic50 binding affinity values. Users may filter on the percentile rank by using the new
Users now have the option of calculating the reference proteome similarity of their filtered epitopes. For this, the peptide sequence for the remaining variants is mapped to the reference proteome using BLAST. Variants where this yields a hit to a reference proteome are marked accordingly and a
.reference_matchesfile provides more information about the matches. This option can be enabled using the
Users may now use the options
all_class_iiinstead of specific prediction algorithms in order to run all prediction algorithms, all class I prediction algorithms, or all class II prediction algorithms, respectively.
For successful pVACvector runs, we now output a
_results.dna.fafile with the most likely nucleic acid sequence for the predicted vector.
When running pVACseq with a proximal variants VCF we would previously assume that your ran VEP with the
--pickoption and only process the first transcript annotation for a variant. With this update we will now associate the correct transcript for a proximal variant with the matching transcript of the main somatic variant of interest.
pvacseq generate_protein_fastacommand now allows users to provide a proximal variants VCF using the
pvacseq generate_protein_fastacommand now supports multi-sample VCFs. Users may use the
--sample-nameto provide the sample name of the sample they wish to process.
pVACseq and pVACfuse would previously error out if the intermediate TSV parsed from the input was empty. In 2.0 the tool will no longer error out but exit with an appropriate message.
pVACvector would previously error out when no valid path was found. In 2.0 pVACvector will not longer error out but exit with an appropriate message.
We now set consistent file permissions on all output files.
We’ve updated our license to BSD 3-Cause Clear. Please note that the individual licenses of our dependent tools remain in place. These can be viewed by on the Tools Used By pVACtools page.
Past release notes can be found on our Release Notes page.
To stay up-to-date on the latest pVACtools releases please join our Mailing List.
Jasreet Hundal , Susanna Kiwala , Joshua McMichael, Chris Miller, Huiming Xia, Alex Wollam, Conner Liu, Sidi Zhao, Yang-Yang Feng, Aaron Graubert, Amber Wollam, Jonas Neichin, Megan Neveau, Jason Walker, William Gillanders, Elaine Mardis, Obi Griffith, Malachi Griffith. pVACtools: A Computational Toolkit to Identify and Visualize Cancer Neoantigens. Cancer Immunology Research. 2020 Mar;8(3):409-420. doi: 10.1158/2326-6066.CIR-19-0401. PMID: 31907209.
Jasreet Hundal, Susanna Kiwala, Yang-Yang Feng, Connor J. Liu, Ramaswamy Govindan, William C. Chapman, Ravindra Uppaluri, S. Joshua Swamidass, Obi L. Griffith, Elaine R. Mardis, and Malachi Griffith. Accounting for proximal variants improves neoantigen prediction. Nature Genetics. 2018, DOI: 10.1038/s41588-018-0283-9. PMID: 30510237.
Jasreet Hundal, Beatriz M. Carreno, Allegra A. Petti, Gerald P. Linette, Obi L. Griffith, Elaine R. Mardis, and Malachi Griffith. pVACseq: A genome-guided in silico approach to identifying tumor neoantigens. Genome Medicine. 2016, 8:11, DOI: 10.1186/s13073-016-0264-5. PMID: 26825632.
The pVACtools source code is available in GitHub.
This project is licensed under BSD 3-Clause Clear License.