Skip to content

Latest commit

 

History

History
92 lines (60 loc) · 4.57 KB

CHANGELOG.md

File metadata and controls

92 lines (60 loc) · 4.57 KB

nf-core/ampliseq: Changelog

The format is based on Keep a Changelog and this project adheres to Semantic Versioning.

nf-core/ampliseq version 1.2.0 "Teal Bronze Lion" - 2021

Added

  • #106 - Added support for PacBio data
  • Added --taxon_reference to be able to support both 'silva' and 'unite'
  • #157 - Added possibility to run double cutadapt steps, --double_primer
  • #211 - Added quality filter option --maxEE

Fixed

  • #182 - Fix input in case there are no underscores in sample IDs
  • #186 - Update github actions
  • #187 - Sample ids are in incorrect order in feature-table from PacBio data
  • #201 - Template update for nf-core/tools version 1.12.1
  • #147 - Split make_classifier in two different processes that can be allocated different resources
  • #183 - Don't fetch taxonomy/create classifier when run with --skip_taxonomy
  • #180 - MultiQC, cutadapt and fastQC now work with --multipleSequencingRuns

Dependencies

  • Updated from cutadapt v2.6 to v2.8

Deprecated

nf-core/ampliseq version 1.1.3 - 2020

Added

  • #170 - Cite paper for initial release
  • #111 - Added parameter for user specified manifest file
  • #118 - Added social preview images
  • #135 - Added --trunc_rmin to make sure that auto trunc cutoff retaines a certain fraction of reads

Fixed

  • #172 - Template update for nf-core/tools v1.11
  • #163 - Template update for nf-core/tools v1.10.2
  • #136 - Pipeline fails with remote working directory
  • #152 - Don't fetch taxonomy/create classifier when run with --onlyDenoising

Dependencies

  • Updated from MultiQC v1.6 to v1.9

Deprecated

  • --reads is replaced by --input due to nf-core/tools v1.10.2

nf-core/ampliseq version 1.1.2 - 2019

  • No further changes, except a bugfix for the timezone issue found by @marchoeppner
  • Specification of --qiime_timezone might be required to run the analysis appropriately

nf-core/ampliseq version 1.1.1 - 2019

Pipeline Updates

  • Update from QIIME2 v2018.6 to v2019.10, including DADA2 v1.6 to DADA2 v1.10
  • Export absolute abundance files into 'results/abundance-table/filtered/' for optional external secondary analysis

Bugfixes

  • #78 - All sequenced classifed to the same species

nf-core/ampliseq version 1.1.0 "Silver Lime Bee" - 2019

Pipeline updates

  • #40 - Added support for data originating from multiple sequencing runs
  • #53 - DADA2 report is always exported
  • #49 - Allowed more filtering options
  • #5 - Introduced check for existence of input files
  • Extended parameter sanity check, including #15
  • #61 - Improved documentation
  • #62 - Utilize nf-core/configs centrally for this pipeline
  • #63 - QIIME imports files by using a manifest, giving more freedom with input file names
  • #84 - Add proper nf-core logo

Bug fixes

  • #57 - Indicate exact regex for sequencing file names
  • #60 - publish demux.qza when --untilQ2import

nf-core/ampliseq version 1.0.0 "Olive Steel Panda" - 2018-11-23

Initial release of nf-core/ampliseq, created with the nf-core template.