Recommendations for the formatting of Variant Call Format (VCF) files to make plant genotyping data FAIR

Abstract

In this opinion article, we discuss the formatting of files from (plant) genotyping studies, in particular the formatting of metadata in Variant Call Format (VCF) files. The flexibility of the VCF format specification facilitates its use as a generic interchange format across domains but can lead to inconsistency between files in the presentation of metadata. To enable fully autonomous machine actionable data flow, generic elements need to be further specified. We strongly support the merits of the FAIR principles and see the need to facilitate them also through technical implementation specifications. They form a basis for the proposed VCF extensions here. We have learned from the existing application of VCF that the definition of relevant metadata using controlled standards, vocabulary and the consistent use of cross-references via resolvable identifiers (machine-readable) are particularly necessary and propose their encoding. VCF is an established standard for the exchange and publication of genotyping data. Other data formats are also used to capture variant data (for example, the HapMap and the gVCF formats), but none currently have the reach of VCF. For the sake of simplicity, we will only discuss VCF and our recommendations for its use, but these recommendations could also be applied to gVCF. However, the part of the VCF standard relating to metadata (as opposed to the actual variant calls) defines a syntactic format but no vocabulary, unique identifier or recommended content. In practice, often only sparse descriptive metadata is included. When descriptive metadata is provided, proprietary metadata fields are frequently added that have not been agreed upon within the community which may limit long-term and comprehensive interoperability. To address this, we propose recommendations for supplying and encoding metadata, focusing on use cases from plant sciences. We expect there to be overlap, but also divergence, with the needs of other domains.

DivBrowse – interactive visualization and exploratory data analysis of variant call matrices

“…These indeterminacies have made fully automated processing difficult or impossible without manual intervention in terms of interoperability at the level of bioinformatics toolchains. As DivBrowse continues to evolve, we look forward to incorporating recent advances and improvements related to metadata in VCF files to better take advantage of the FAIR data paradigm, such as improved interoperability and reusability [25].…”

“…For example, the usage of standardized sample-IDs for the genotypes based on BioSamples-IDs in the VCF metadata section would allow DivBrowse to easily read those BioSamples-IDs and use them to automatically interconnect the genotypes in the GUI of DivBrowse with other omics-related web-based information systems [26]. As another example, the appropriate standardized usage of the VCF metadata field "Contig", as recommended in Beier et al [25],…”

“…For example, the usage of standardized sample-IDs for the genotypes based on BioSamples-IDs in the VCF metadata section would allow DivBrowse to easily read those BioSamples-IDs and use them to automatically interconnect the genotypes in the GUI of DivBrowse with other omics-related web-based information systems [26]. As another example, the appropriate standardized usage of the VCF metadata field “Contig” , as recommended in Beier et al [25], would allow to automatically derive human-readable chromosome labels, the chromosome length and the species from the VCF file. This would make corresponding manual entries for the chromosome labels and length in the DivBrowse configuration file unnecessary in future.…”

“…The Variant Call Format (VCF), that acts as the input file format for genotypic data for DivBrowse, has suffered from a number of indeterminacies in the past with regard to the FAIR criteria [33,34]. These indeterminacies have made fully automated processing difficult or impossible without manual intervention in terms of interoperability at the level of bioinformatics toolchains.…”

“…It thus provides a reusable tool to efficiently deliver the increasing amount of genotypic data to the public in the form of an interactive visualization with easy to use basic analysis functions and multiple entry points. The Variant Call Format (VCF), that acts as the input file format for genotypic data for DivBrowse, has suffered from a number of indeterminacies in the past with regard to the FAIR criteria [33,34]. These indeterminacies have made fully automated processing difficult or impossible without manual intervention in terms of interoperability at the level of bioinformatics toolchains.…”