Skip to content

Latest commit

 

History

History
164 lines (90 loc) · 5.79 KB

CHANGELOG.md

File metadata and controls

164 lines (90 loc) · 5.79 KB
Raw

Atria Change Log

TODO

  • Feature: multiple primer trimming.
  • Feature: UMI trimming.

v4.1.1

  • Fix: -z NUM -Z NUM error when length to trim < 0.

v4.1.0

  • Change: --length-range default change from 50:500 to 30:999999.
  • Feature: HardClipEnd: new process to hard remove the last N bases.
  • Change: names in processing order (--order -O) changed.
  • Feature: PCRDedup: remove PCR duplicates from fastq files. The entire paired sequence is compared and hashed. This method require large memory because it stores hashes of reads. To enable, use --pcr-dedup.
  • Feature: processing stats are recorded in the json file.
  • Fix: polyX_tail_scan algorithm now is more precise, and tailing Ns also count.

v4.0.3

  • Fix: --order or -O option should accept multiple arguments.

v4.0.2

  • Fix: --detect-adapter for paired reads: refer to index 1 of empty vector when no adapter is found.

v4.0.1

  • Fix: dep cihga39871/BioSequences.jl: detailed error message if input files' line break is '\r\n'.

v4.0.0

  • Optimize: algorithm: now the non-overtrim rate for reads without adapters are higher.
  • Feature: re-write trimming to allow trim multiple adapters at the same time. This change is adjusted for metabarcoding data.
  • Feature: hard-clip: now hard-clip arguments do differently for r1 and r2. This change is adjusted for metabarcoding data. Remove -C --clip-after -c --clip5; add -b --clip-after-r1 -B --clip-after-r2 -e --clip5-r1 -E --clip5-r2.
  • Optimize: --detect-adapter for paired-end reads now guess adapters from pair information, rather than the existing adapter pool.
  • Feature: users can customize order of processing: -O | --order.

v3.2.2-1

  • Fix: undef error of is_concensused when enabling --stat (thanks to kalavattam, #3)

v3.2.2

  • Optimize: speed up for threads <= 2.
  • Fix: atria test should not depend on source files.

v3.2.1

  • Feature: automatically skip completed analyses. Use --force or -f to disable the feature.

v3.2.0

  • Remove multi-proc mode since it is unstable.

v3.1.4

  • Logging: new logging for versions and sample completion.
  • Fix v3.1.3: multi-proc mode: Julia v1.8.1 does not allow assign new ARGS, and add -t nthread in julia_args.
  • Fix v3.1.3: pe-consensus: error when insert_size = -1; fix trimming when insert_size = -1.
  • Benchmark iscomplement in Atria v3.1.2 and that in BioSequences, and found it is good to stick to BioSequences.

v3.1.3

  • Compatible: Julia v1.8 and BioSequences v3.1.0.
  • Fix: quality offset not changed in some places when providing a different --quality-format.
  • Fix: use Base.invokelatest to bypass world age for functions evaluated at run time.
  • Docs: update.

v3.1.2

  • Fix: optimize output file names if ending with .bz2.

v3.1.1

  • Fix: when reporting an encode error, report the previous and current lines instead of the whole chunk of data.

v3.1.0

  • New feature: --detect-adapter for adapter determination.

  • Fix: when input is an empty compressed fastq, atria exits with error because read_chunks!(::IO, ...) should return 4 elements, but returned 2.

v3.0.3

  • Fix v3.0.2: will_eof should be true when unknown.

  • Do not resize chunk sizes before cycle 1 when inputs are compressed and cannot determine uncompressed sizes. Just assume data are not trimmed before.

v3.0.2

  • Fix uncompressed_size1 not defined on gzipped single-end input (#2).

v3.0.1

  • Avoid to lock IOStream when write fastq in thread_output.jl: replace write(::IOStream, ...) with write_no_lock(::IOStream, ...). It is slightly faster.

  • Speed optimization for consensus calling: overwrite BioSequences.complement(::DNA) (1.40X), and define iscomplement(::DNA, ::DNA) (1.79X).

  • Other minor parallel implementations.

v3.0.0

  • If users choose to trim adapter, check 1 bp offset of adapter sequences. It is because Atria might have 1 bp error in some cases.

v2.1.2

  • Parameter optimization using atria simulate: --trim-score-pe 19->10, --tail-length 8->12.

  • Development of Atria simulation methods.

v2.1.1

  • Fixing wrapper_single_end.jl: cannot trim true adapter position at index of -1.

v2.1.0

  • If a r1/2 adapter is found, but the region of r2/1 is missing or its quality too low (mean prob < 0.6), skip PE check and just trim like single-end. With this, trim_score do not need to compensate for the situation, so rise the default trim-score-pe (10->19).

v2.0.0

  • Supporting low-complexity filtration.

  • Supporting polyX tail trimming.

  • Supporting single-end fastq.

  • Supporting bzip2 compression/decompression.

  • Supporting non standardized gzip compression files.

  • Optimizing default parameters. (r1-r2-diff 0->0, trim-score-pe 8->10, score-diff removed, kmer-n-match 8->9)

  • Robustness optimization: the lower bound of match probability is set to 0.75 because match probability lower than 0.75 is outlier and affect trim score strongly.

v1.1.1

  • Performance optimization: adapter and PE trimming: following v1.1.0-1, if the loosen match's nmatch > trim_score, replace the old one.

v1.1.0

  • Performance optimization: adapter and PE trimming: if no adapters were matched, the number of errors of PE match is loosen.

  • Performance optimization: consensus calling: new arg --kmer-tolerance-consensus 2->10; optimized arg --min-ratio-mismatch 0.2->0.28.

  • Speed optimization: check overlap_score > 0 before computing score (pe_consensus!).

v1.0.3

  • More detailed error output when encoding a non-nucleotide character (throw_encode_error(...)).

  • Following symbolic link before checking file size for non-Windows platforms (check_filesize(::String)).

  • When run in multi-file parallel mode, write stdout and stderr to a 'stdlog' file (julia_wrapper_atria(...)).

  • Add option --check-identifier to check whether the identifiers of r1 and r2 are the same.

v1.0.2

  • First mature version of Atria.