v2.2.0

Added

• Output {{sample}}.stats.json file describing some key metrics for the analysis.
• Summary of gene coverage if a 4-column BED is provided.
• Automated sex determination using relative coverage of chrX and chrY.
• Retry strategy added to snp:aggregate_pileup_variants to prevent out of memory error.

Changed

• --GVCF --phased will produce a phased GVCF.
• Changed default phasing algorithm to whatshap, with the possibility to change the phasing to longphase with --use_longphase true.
  • The intermediate phasing is still performed using longphase.
• Setting --snp --sv --phased will emit individually phased SNPs and SVs.
• Phased bedMethyl files now follow the pattern {{ alias }}.wf_mods.{{ haplotype }}.bedmethyl.gz.
• --sex parameter uses XX and XY rather than "female" and "male".
• Update modkit to v0.2.6.
• Improved modkit runtime by increasing default threads and increasing the default interval size.
• Improved modkit runtime by increasing the default interval size and running modkit on individual contigs.
• modkit is now run only on chromosomes 1-22, X, Y and MT, unless --include_all_ctgs is provided.
• Increased minimum CPU requirement for the workflow to 16.
• Filtering of SVs using a BED file now includes sites only partially overlapping the specified regions.
• basecaller_cfg will be inferred from the basecall_model DS key of input read groups, if available
  • Providing --basecaller_cfg will not be required if basecall_model is present in the DS tag of the read groups of the input BAM
  • --basecaller_cfg will be ignored if a basecall_model is found in the input BAM
• Reconciled workflow with wf-template v5.1.2
• Update to Clair3 v1.0.8.
• Update to longphase v1.7.1.

Fixed

• Update schema to allow selection of multiple BAM files per sample in EPI2ME.
• Spectre CNV report not handling cases when no CNVs detected.
• Lines denoting normal maximum and pathogenic minimum thresholds now correctly displayed on STR repeat content plots.
• Workflow will not emit sample.cram if sample.haplotagged.cram has been created by the workflow to save storage.
• Emitting nonsense input.1 file

Removed

• Single-step joint phasing of SV and SNP.
• --output_separate_phased as the workflow emits only individually phased VCFs.
• A copy of the reference and the generated reference cache is no longer output by the workflow.
  • The workflow encourages use of readily available standard reference sequences, so re-emitting the input reference as a workflow output is unnecessarily consuming disk space.

v2.1.0

Changed

• ClinVar version in SnpEff container updated to version 20240307.
• Convert to BAM only when --cnv --use_qdnaseq is selected.
• Update to Clair3 v1.0.6.
• Update Spectre to fix an error when parsing Clair3 VCFs with multiple AFs.
• Support for an input folder of multiple BAM files per sample with --bam (instead of only allowing a single BAM per sample).
• refine_with_sv to be run by chromosome in order to reduce memory footprint.

Fixed

• Force minimap2 to clean up memory more aggressively. Empirically this reduces peak-memory use over the course of execution.
• Handling of input VCF files with --vcf_fn.
• --phased --sv --snp generates a truncated VCF file when # appears in the VCF INFO field
• Some reporting scripts using too much memory.

Removed

• CRAM as supported input format.
• old_ref parameter as providing the reference of an existing CRAM is no longer needed.

v2.0.0

Changed

• CNV calling with --cnv is now performed using Spectre, which is optimised for long reads.
  • Legacy CNV calling using QDNAseq may still be carried out with --cnv --use_qdnaseq.
  • The bin size parameter has been renamed from --bin_size to --qdnaseq_bin_size.
• Skip CNV CRAM to BAM conversion if downsampling is required, to avoid creating an unnecessary intermediate file.
• The output of --depth_intervals now has .bedgraph.gz extension.
• SV workflow outputs SVs in the autosomes, sex chromosomes and MT; use --include_all_ctgs to output calls on all the sequences.

Added

• Output definitions for coverage files.
• N50 and mean coverage added to alignment report.

Fixed

• EPI2ME Desktop incorrectly allowed selection of directory for tr_bed.
• failedQCReport failing to generate a report.

v1.11.0

Changed

• Add an additional whatshap haplotag process after the final VCF phasing.
• Updates to the phasing subworkflow significantly impact the runtime and storage requirement for the workflow, as detailed here.
• Several performance improvements which should noticeably reduce the running time of the workflow

Fixed

• Updated the version of Straglr, which addresses the following:
  • Repeats can now be called in RFC1
  • Start position of called STRs is 1-based rather than 0-based
  • VCF headers now match those in the FORMAT field
• Generate allChromosomes.bed using samtools faidx index instead of pyfaidx, to avoid a KeyError
• Inconsistent file ownership of bundled Clair3 model files which could lead to subuid errors in some environments

v1.10.1

Fixed

• Bug report form.

v1.10.0

Added

• Clair3 4.3.0 models.

Changed

• --phase_vcf, --joint_phasing and --phase_mod are now deprecated for --phased; see the README for more details.
• --use_longphase_intermediate is now deprecated, and --use_longphase false will use whatshap throughout consistently
• Running --phase --snp --use_longphase false will now phase indels too
• --basecalling_cfg currently provides the configuration to Clair3.
• The clair3: prefix to Clair3 specific models is no longer required.

Fixed

• CNV report generation fails if there is no consensus on the copy number of a chromosome
  • Undetermined category has been added to the Chromosome Copy Summary to account for these cases
• readStats reports metrics on the downsampled BAM when --downsample_coverage is requested.
• Spurious warning for missing MM/ML tags when a BAM fails the coverage threshold

Removed

• wf-basecalling subworkflow
  • fast5_dir input and other basecalling related options have been removed from the workflow parameters
  • Users should run the standalone wf-basecalling workflow and provide the output to wf-human-variation
• Mapula statistics with --mapula

v1.9.2

Fixed

• --joint_phasing generating single-chromosome VCF files.

v1.9.1

Changed

• ClinVar annotation of SVs has been temporarily removed due to not being correctly incorporated. SnpEff annotations are still produced as part of the final SV VCF.
• New documentation

Removed

• --annotation_threads parameter, as the SnpEff process does not support multithreading.

Fixed

• Truncated SV VCF header generated from vcfsort.
• sed crashing with I/O error in some instances.
• Missing flagstats file in output directory.

v1.9.0

Added

• STR workflow report now includes additional plots which display repeat units and interruptions in each supporting read
• CNV workflow now outputs an indexed VCF file to the output directory

Changed

• Legend symbols in STR genotpying plot
• Unambiguous naming of bedMethyl files generated with --mod
  • Unphased outputs will have the pattern [sample_name].wf_mods.bedmethyl.gz
  • Phased outputs will have the pattern [sample_name]_[1|2|ungrouped].wf_mods.bedmethyl.gz

Fixed

• Report step failing if bcftools stats file has only some sub-sections
• Clair3 ignoring the bed file
• merge_haplotagged_contigs incorrectly generating intermediate CRAM when input is BAM
• STR content generation failing due to forward slash in disease name in variant_catalog_hg38.json
• Report name for the read alignment statistics now follows the pattern [sample_name].wf-human-alignment-report.html

v1.8.3

Fixed

• configure-jbrowse breaking on unescaped spaces