Changelog¶

v2.0¶

v2.0 is a major rewrite of snpArcher with breaking changes to the sample sheet schema, config structure, and module interfaces.

Breaking changes¶

Sample sheet schema redesigned. The v1 columns (BioSample, Run, LibraryName, fq1, fq2, refGenome, refPath, SampleType, lat, long) are removed. v2 uses sample_id, input_type, input, with optional library_id and mark_duplicates.
Reference genome moved to config. refGenome and refPath are no longer sample sheet columns. Use reference.name and reference.source in config.yaml.
Sample metadata split into a separate file. Columns like exclude, lat, and long are now in an optional metadata CSV specified by the sample_metadata config key.
Config structure is now hierarchical. Flat v1 keys (e.g., minNmer, cov_filter, mappability_k) are replaced by nested keys (e.g., intervals.min_nmer, callable_sites.coverage.enabled, callable_sites.mappability.kmer).
Module config is nested under modules. QC and postprocess options are now under modules.qc.*, modules.postprocess.*, etc.
v1 config keys trigger a hard error. If legacy v1 keys (e.g., final_prefix, refGenome) are detected, the workflow exits with an error message.

Sample sheet column mapping (v1 to v2)¶

v1 column	v2 column	Notes
`BioSample`	`sample_id`	Must match `^[A-Za-z0-9._-]+$`.
`Run`	`input`	Set `input_type: srr`.
`fq1`, `fq2`	`input`	Join as `fq1;fq2`, set `input_type: fastq`.
`LibraryName`	`library_id`	Optional; defaults to `sample_id`.
`refGenome`	`reference.name` (config)	Moved to config.
`refPath`	`reference.source` (config)	Moved to config.
`SampleType`	`exclude` (metadata sheet)	`exclude: true` in metadata replaces `SampleType: exclude`.
`lat`	`lat` (metadata sheet)	Moved to optional metadata CSV.
`long`	`long` (metadata sheet)	Moved to optional metadata CSV.
`mark_duplicates`	`mark_duplicates`	Now a per-row sample sheet column or global config default.

Config key mapping (v1 to v2)¶

v1 key	v2 key
`refGenome`	`reference.name`
`refPath`	`reference.source`
`sentieon`	`variant_calling.tool: sentieon`
`sentieon_lic`	`variant_calling.sentieon.license`
`minNmer`	`intervals.min_nmer`
`cov_filter`	`callable_sites.coverage.enabled`
`mappability_k`	`callable_sites.mappability.kmer`
`mappability_min`	`callable_sites.mappability.min_score`
`mappability_merge`	`callable_sites.mappability.merge_distance`

New features¶

New input types: bam and gvcf input types allow entering the pipeline at later stages.
Callable sites overhaul: Coverage analysis uses mosdepth D4 and Zarr-based processing. Auto-thresholding computes min/max coverage from the depth distribution.
Interval-based parallelization for GATK: Hierarchical gVCF/VCF concatenation with configurable batch sizes.
Parabricks support: GPU-accelerated variant calling via NVIDIA Parabricks.
DeepVariant support: Deep learning variant caller with GLnexus joint genotyping.
bcftools pathway: Region-parallelized bcftools mpileup/call as an alternative to GATK.

Deprecated aliases¶

These keys are accepted with a warning but should be updated:

Deprecated key	Replacement
`reference.path`	`reference.source`
`variant_calling.gatk.ploidy`	`variant_calling.ploidy`

v1.x¶

The original snpArcher release described in:

Mirchandani et al. (2024). "A fast, reproducible, high-throughput variant calling workflow for population genomics." Molecular Biology and Evolution, 41(1), msad270. doi:10.1093/molbev/msad270

v1 supported GATK and Sentieon variant calling with a flat config structure and a single sample sheet combining sample data, reference paths, and metadata. See the MBE paper for benchmarking results and design rationale.