GATK (Genome Analysis Toolkit)

gcnvkernel is a Python module that serves as the computational backend for GATK germline copy number variant (gCNV) tools and workflows. It implements inference schemes for read-depth profile denoising, germline integer copy number variation discovery, and germline contig ploidy determination.

While intended to be called privately by GATK CLI tools (where its outputs are treated as intermediate data), advanced users can use it directly via provided Python CLI scripts or in interactive Python environments for exploratory analysis.

Funcotator (FUNCtional annOTATOR) is a functional annotation tool that analyzes variants and produces results in VCF (with INFO field annotations) or MAF formats. It uses data sources like GENCODE, dbSNP, gnomAD, and COSMIC to classify variants and provide gene information.

M2 Integration Details:

• By default, the M2 WDL runs Funcotator to produce a TCGA MAF from the M2 VCF.
• Metadata: You should pass specific parameters to populate TCGA MAF metadata fields. These are ignored if Funcotator is not run.
• Limitations: Certain TCGA MAF fields (like validation alleles) cannot be generated by M2/Funcotator and must be handled by downstream processes.
• Vocabulary: Funcotator does not strictly enforce the TCGA MAF controlled vocabulary; users are responsible for correct specification. Consequently, generated MAFs may not pass the TCGA Validator.

Funcotator (FUNCtional annOTATOR) is a functional annotation tool within the GATK toolset designed for both somatic and germline use cases. It analyzes variants in a VCF file and labels them with one of twenty-three distinct variant classifications.

Key capabilities include:

• Producing gene information (e.g., affected gene, predicted amino acid sequence).
• Associating variants with data sources like GENCODE, dbSNP, gnomAD, and COSMIC.
• Outputting results in either Variant Call Format (VCF) or Mutation Annotation Format (MAF).
• Allowing user-defined annotations via extensible and configurable data sources without modifying Java code.

Python packages used by GATK must be placed in this directory, with each package contained in its own subdirectory.

• Standalone Packages: If a package can be installed independently, you may place a setup_<PACKAGE_NAME>.py file in this directory.
• Standard GATK Installation: During the creation of the common GATK conda environment, setup.py combines all packages and pip-installs them as a single package named gatkpythonpackages.

The --transcript-selection-mode parameter determines how the primary annotated transcript is selected. The available modes are:

• CANONICAL (Default): Selects a transcript based on canonical order (User custom list > Curation level > Variant classification rank > Appris annotation > Sequence length > Alphabetical).
• BEST_EFFECT: Selects a transcript based on functional effect (User custom list > Variant classification rank > Curation level > Appris annotation > Sequence length > Alphabetical).
• ALL: Similar to CANONICAL, but does not drop any transcripts; it renders all overlapping transcripts.

When working with Funcotator or GATK, it is critical to identify which human reference genome version is being used. Differences between versions like GRCh37, hg19, b37, and HumanG1Kv37 are identified by comparing the MD5sums of their contigs.

In the reference comparison documentation:

• Contigs with identical MD5sums across references are listed in the table rows.
• A blank entry (----) for a contig indicates that the MD5sum does not match, signifying a sequence difference between those specific references.
• Primary contigs with differing MD5sums are highlighted in red.
• Alternate contigs with differing MD5sums are highlighted in orange.

Funcotator uses data sources organized into a specific directory hierarchy. Each main data source folder must contain sub-directories for each supported reference (e.g., hg19, hg38). Inside each reference-specific directory, a .config file is required to detail how to match the data to a variant.

Directory Structure Example:

    dataSourcesFolder/
         Data_Source_1/
             hg19
                 data_source_1.config
                 data_source_1.data.file.one
                 ...
             hg38
                 data_source_1.config
                 data_source_1.data.file.one
                 ...
         Data_Source_2/
             hg19
                 ...

Constraints:

• Multiple data source folders can be specified as input.
• No two data sources can have the same name.

You can add custom annotations by creating a new sub-directory in your main data sources folder. This directory must contain reference-specific folders (e.g., hg38) containing the data files and a mandatory .config file.

Supported data formats include:

• simpleXSV: Separated value table (CSV/TSV) keyed by Gene Name or Transcript ID.
• locatableXSV: Separated value table keyed by genome location (requires contig, start, and stop columns).
• gencode: GENCODE data files (GTF format).
• cosmic: COSMIC data.
• vcf: Variant Call Format files.
• Cloud Sources: You can also point src_file to a cloud URL (e.g., gs://...) in your config file to avoid local storage.