Changelog #

v0.9.0 #

New step: ValidateReadNamesPrintable
New step: CompareStringTags
New step: ConcatTags - concatenate multiple tags into one
New step: Lowercase -unified replacement for LowercaseTag/LowercaseSequence
New step: ValidateReadPairing - Hamming-distance-based read pairing validation (This is the only automatically injected step in fastqrab!)
New Step: ConvertToRate
New step: ExtractIUPACSuffix
New step: FillMissing
- to, for example, choose ’either’ found barcode.
New step: Matches to restrict to a debug selection of reads.
New step: CalcWorstQuality
New step AssignByHalves: for Cell Ranger like probe assignment
New step: StoreSingleCellMatrix to quickly calculate single cell matrix from a cell barcodes, assigned genes and UMIs.
New step: StoreTagInSequence, allows storing at another tags location (contrast with StoreTagBackInSequence)

ExtractAnchor merged into ExtractRegions
NCount renamed CalcNContent to be in-line with CalcGCContent and CalcBaseContent
TagOtherFile unifies OtherFileByName and OtherFileBySequence into one step
‘StoreTagInSequence’ -> ‘StoreTagBackInSequence’

Conditional Swap/ReverseComplement variants merged into the main steps
if tag support on 8 core editing steps for conditional read editing
min_length added to ExtractRegions
(optional) Quality checking added to Prefix / Postfix
Tag value replacement within regular expressions on ExtractRegex
Swap - Fixed a correctness bug.
Progress stdout handling unified, reports core configuration and time taken by finalization of steps
ExtractRegex from tags
StoreTagsInTable - optionally do not output read names.
StoreTagInComment now supports multiple tags at once
ReverseComplement can now be applied to tags
CalcBaseContent outputs 0..=1, not 0..=100
More steps support ‘relative’ as an option (CalcGCContent, CalcNContent, CalcQualifiedBases)

Multiple queries in one step, max_mismatches now required, large performance improvements
Tie handling for multiple queries
Can now also search within a (small) distance from start/end.

Can now assign either barcodes or barcode labels
Uses the hamming-resonate crate.
Gained support for explicit on_no_match and on_tie handling in multiple variations, with or without name-prefix based equivalence classes.
multi-core
on_tie=ByEditProbability supports reading the barcode distribution from a previously generated report.

Tag histogram reports; demultiplex data nested under ‘demultiplex’ key in reports

Store tags in BAM tags
Store one tag in BAM reference field.
Corrected error message when read length > max bam read length
better error messages when the read names can’t be represented in BAM. (see also ValidateReadNamesPrintable

Redesigned multi-core engine: workpool based, better controllable, better documented
Default thread count now uses all available CPU cores
Rapidgzip for parallel gzip decompression, now also for FASTA; auto-detected; included in Nix builds
Arena-based parsers for FASTA and BAM
Parallel BAM decoding
Multicore EvalExpression, ReportCountOligos, ReportLengthDistribution
Prefix/Postfix massively improved performance
Merge base statistics ~80% faster
ConcatTags ~15% faster
IUPAC matching: replaced Sassy with optimized pure-Rust implementation
Optimized SwapConditional, TrimAtTag, StoreTagBackInSequence, FilterReservoirSample, Rename
Dynamic cuckoo filter sizing; initial_filter_capacity documented; read count estimation
ExtractLongestPolyX no longer O(n^2)

verify command: validates a pipeline produces expected output; auto-detects config, captures stdout/stderr
configuration TOML can now be read from stdin (incompatible with reads from stdin).
barcode sections can now read from files (FASTA, txt)
Shell autocompletion for bash, fish, and zsh
benchmark mode and per-step benchmark harness
template command: shows help on error
LLM configuration guide and template.toml rewrite for LLM-assisted config generation
TagLabel type: all tag names are now strongly typed; duplicate tag names produce a clear error
IndexMap replaces HashMap everywhere for deterministic output order
unwrap() replaced with expect() throughout; clippy::unwrap_used now denied
MSRV pinned to match flake.nix Rust version
Security: upgraded bytes crate (GHSA-434x-w66g-qw3r)
Upgraded dependencies
Reports: Prevent script injection.
Report’s count_oligos now supports named oligos.

Fixed fastp merge algorithm (replaced with direct port of the reference C++ algorithm)
Fixed invalid FASTQ detection when comment line doesn’t start with ‘+’
Fixed Windows newline detection edge case in parser
Fixed Local-Local FastQElement swap
Fixed demultiplex & fragment count in reports
Fixed Head short-circuit (broken by SpotCheckReadPairing)
Fixed ignore_unaligned → now include_mapped / include_unmapped
Fixed barcode overlapping multiple matches
Fixed a bug n [ExtractIUPACSuffix](/fastqrab/main/docs/reference/tag-steps/extract/ExtractIUPACSuffix/
Fixed a potential panic in ExtractLongestPolyX