Changelog

v.0.4

v 0.4.0

Overview

This release is primarily an internal refactoring focused on improved robustness: - large source files are split into focused modules.
- almost all methods and functions have been refactored significantly for clarity, maintainability and robustness.
- test coverage is expanded to >90% for most code.
- there are also several small bug fixes and performance improvements.

User-facing changes

aggregateFile replaced by aggdb

The main user-facing change is the replacement of aggregateFile with a new SQLite-backed aggdb class.

The aggregateFile class is replaced by aggdb, an SQLite-backed class (extending SQLiteConnection, matching the gdb pattern). This provides fast queries via SQL instead of sequential file scanning. Related methods:

• collapseAggregateFiles -> collapseAggDbs
• mergeAggregateFiles -> mergeAggDbs
• aggregateFileList -> aggdbList

The aggdb stores results in structured tables (meta, params, SM, aggregates) and supports efficient merging.

Other user-facing updates

• writeVcf, spatialClust: new memlimit parameter for batch processing, improving performance on large datasets.
• concatGdb, uploadAnno, uploadCohort: new overWrite parameter (defaults to FALSE). Existing code that relied on implicit overwrites now needs to pass overWrite = TRUE.
• concatGdb: targets now accepts a character vector in addition to a file path. New sample-identity validation across input gdbs.
• rvatViewer: Shiny app removed (will be moved to a separate package soon).
• skat_fwe / skat_burden_fwe / skato_fwe tests removed. These can be replaced by the resampling functionality already present in assocTest.

v.0.3

v.0.3.2

• Bug fix: Fixed a bug in the getAnno method: if the ranges parameter was specified, but none of the variants in the specified table overlap with those respective ranges, all rows in the specified table were returned.

v.0.3.1

• Includes new mutationPlot function to create mutation plots that visualize variant-level and gene-level association results along a transcript structure. Optionally, custom tracks such as protein domains or mutation clusters can be overlaid.
  
• The mutationPlot function is demonstrated in the updated spatial clustering tutorial.
  
• Includes a new method getRanges for varSetList and varSetFile that retrieves genomic ranges for variant sets. Can be used to retrieve either genomic coordinates or other user-defined coordinates such as CDS coordinates genearated with the mapCDS method.

v.0.3.0

Large update! In particular, changes have been made to:

• Documentation and tutorials
  • All methods/classes/functions now contain examples in their documentation.
  • New tutorials are included on variant annotation, gene set analysis and spatial clustering.
    Existing workflows have been updated with new features/functions .
• Association testing
  • RVAT results now include a OR field indicating the odds-ratio for logistic tests (firth and glm). Importantly: For firth and flm tests effect/effectSE/effectCIlower/effectCIupper are now on the log-scale!
  • Effect alleles (i.e. the allele to which the effect estimate refers) and non-effect alleles are now included in the results of single variant tests.
• Metadata
  • GDB now includes additional metadata, including a unique identifier, the genome build and creation date and can be retrieved using corresponding getters (e.g. getGenomeBuild and getGdbId). The genome build can be set when building the gdb (genomeBuild parameter), and will then be used correctly assign ploidies on the sex-chromosomes in downstream analyses (so no need to set the checkPloidy parameter anymore, although this is still an option). The gdb identifier can be useful to track which gdb was used to generate results, as the id will be included in the metadata of downstream files such as results files and varSetFiles (see below). Also, it is used methods such as assocTest and summariseGeno to check whether supplied varSetFile/varSetList files were generated from the specified gdb.
  • Files generated using RVAT, such as rvatResult and varSetFiles, now include a header with metadata, including the RVAT version used to generate the file, the gdb identifier, the genome build and the creation date. This metadata will be included when loading or connecting to these files with the corresponding RVAT methods (e.g. rvbResult(<file>) and varSetFile(<file>)). Note that if you read in RVAT results using non-RVAT functions (e.g. read.table), then you will have to skip lines that start with ‘#’.
• Robustness, fixes & efficiency
  • Unit tests are now included that cover the majority of code.
  • RVAT now passes R CMD CHECK without errors / warnings.
  • Complete rewrite of the RVAT command-line interface. This is mostly an ‘under-the-hood’ update to increase robustness and maintainability. For users it is relevant that 1) CLI + help pages are now in-sync with interactive version 2) parameters can now be provided without an ‘=’, e.g. --gdb <path to gdb> rather than --gdb=<path to gdb>
  • Increased efficiency: significant speed-up of extracting ranges (ranges parameter in getGT and getAnno).
  • Several bug fixes / improvements.
• Miscellanous
  • Annotations can now be included directly when loading genotypes using getGT()by setting the anno parameter, or setting inludeVarInfo = TRUE to include the ‘var’ table. If ‘REF’ and ‘ALT’ alleles are included in the annotations, ‘effectAllele’ and ‘otherAllele’ are assigned in the genoMatrix and automatically updated when alleles are flipped. effectAllele and otherAllele are passed on as a fields in single variant results.
  • subsetGdb: Now has a VAR_id parameter to directly subset VAR_ids and a tables parameter to specify which tables to keep.
  • getCohort now returns records only for the samples included in the respective cohort, rather than running NA fields for all excluded samples. The keepAll parameter can be set to TRUE to return all records, like in previous versions.
  • uploadAnno: now has a keepUnmapped flag, if set to TRUE (default) variants that do not map to gdb are discarded.
  • writeVcf now includes a includeVarId parameter that controls whether VAR_ids should be included in the ‘ID’ field. This parameter defaults to FALSE, in which case the ‘ID’ field from the ‘var’ table is included. Note: this is a change relative to the previous version, in which VAR_ids were written to the ‘ID’ field by default. Also, FORMAT field not included anymore when generating a sites-only vcf (includeGeno=TRUE).
  • ‘AF’ field in rowData slot of genoMatrix is removed.
  • mergeAggregateFiles is now split into two methods: mergeAggregateFiles and collapseAggregateFiles. mergeAggregateFiles behaves identically to former mergeAggregateFiles method with collapse = FALSE, whereas collapseAggregateFiles behaved identically to former mergeAggregateFiles method with collapse = TRUE.