• Important: The output columns have changed: gnomADe_* replaces gnomAD_* per VEP output
• Better support for more recent VEP releases, including updates of test dataset, though YMMV
• Removed --af_esp parameter for VEP since it is deprecated (Thanks @bozbezbozzel and @MatthijsPon)

• Important: The output columns have changed with the removal of ExAC_* data in favor of gnomAD_* data
• Important: FILTER tag common_variant is now applied if any of the gnomAD subpopulation AFs are >0.04%
• New options --vep-custom and --vep-config to operate VEP options --custom and --config respectively
• New option --retain-ann to extract more data from INFO/ANN or INFO/CSQ including VEP custom fields
• New option --vep-overwrite to overwrite a VEP annotated .vep.vcf file if found, instead of erroring out
• New option --verbose to show more status messages during runtime
• Old option --filter-vcf removed in favor of --vep-custom which can do the same thing but better

Credits to @alanhoyle for these new features!

• Partial support for snpEff VCFs when using --inhibit-vep
• maf2vcf: Sanity checks for invalid alleles
• maf2maf: Remove only one anchor bp if found, not all of them
• Fix bug breaking maf2maf when --filter-vcf is undefined
• Updates to dockerfile, dependencies, and tests

• Updates and fixes to isoform lists from MSKCC and Uniprot
• New isoform list with MANE Select isoforms
• New isoform list for MSK users of merged Refseq+Ensembl VEP cache
• Faster Travis CI tests that run offline; Better Dockerfile
• Handle new start_retained_variant variant consequence
• Disable --filter-vcf option by default

• Added arg --inhibit-vep to skip running VEP, but still extract any VEP data if found
• New Dockerfile that uses Clear Linux and conda for simpler and faster VEP/vcf2maf
• Minor fixes to Entrez IDs, isoform overrides, and documentation
• Handle scRNA and lncRNA biotypes
• Add AR gene to uniprot isoform overrides

• Added arg --retain-fmt to vcf2maf, which retains FORMAT fields
• Fixed handling of allele depths in vcf2maf, for Strelka indels
• In maf2vcf, feed samtools faidx 5k regions at a time, not 50k
• Fix: maf2maf is told --filter-vcf=0, but vcf2maf did not know

• Support for newest VEP (v91), while remaining compatible with older VEP (v88 or older).
• vcf2maf/maf2maf now report gnomAD AFs, though they're not yet used for any filtering.
• Option to disable common_variant tagging: Set --filter-vcf 0 in vcf2maf/maf2maf.
• HGVSp_Short will now use = for silent muts. E.g. p.T125= instead of p.T125T.
• vcf2maf will no longer annotate ALTs seen in both tumor and normal GT.
• maf2vcf handles multiallelic sites better.
• maf2vcf generates VCFs that should pass the EBI validator.
• New uniprot isoform list from Ensembl Genes 91 in http://grch37.ensembl.org/biomart

• vcf2maf now supports Delly VCFs, producing MAFs with fields needed by cBioPortal.
• Fix: Remove INFO/SVTYPE when ALT is defined (e.g. Pindel), or else VEP skips transcript-level effect.
• Using --remap-chain now retains original loci in a column named REMAPPED_POS.
• New chain files added for GRCh38_to_GRCh37 and GRCm38_to_NCBIM37.
• New unit tests with variants that are tricky to annotate.

• Fix: Corrected MSK canonical isoforms for four genes as explained in PR #127.
• Fix: Changed comma-delimited FILTER values to semicolon-delimited, per VCF specs.
• maf2vcf adds dummy FILTER descriptions to VCF header, to play nice with GATK and bcftools.
• vcf2vcf retains header lines for source, reference, assembly and phasing, if found.

• Added p14ARF as a canonical isoform for CDKN2A, in addition to p16INK4A.
• vcf2vcf can now run mpileup on tumor/normal BAMs to populate an input VCF with allele counts.
• vcf2vcf can parse/edit INFO/FORMAT fields, and also liftOver variants if given a --remap-chain.
• Added support for cgpPindel and CaVEMan VCFs from Sanger pipelines.
• Added arg --any-allele to allow mismatched variant alleles, when VEP reports co-located variants.
• Added arg --buffer-size to control number of variants loaded at a time, dependent on available RAM.
• Bug Fixed: all_effects column was skipping effects without gene names, like in regulatory regions.
• Other minor fixes and additions.