Deblur

Official version 1.1.0 Released on 12 September 2018.

Bug fixes
* Fixed problem causing deblur to ignore the indel error uper bound and use the mismatch error upper bound instead. This caused deblur to use the mismatch depenent default error rate instead of the default indel error rate (which is constant for up to 3 indels). See [issue 178](https://github.com/biocore/deblur/issues/178) for more details.
* Fixed problem running deblur on new python versions (due to python treating the sparse matrix size (1E9) as float rather than int).

Official version 1.0.4 Released on 24 January 2018.

Bug fixes
* During checking for chimeras via vsearch, lower quality nucleotides were indicated by lower case bases. This is not informative for Deblur, but led to issues on downstream analyses, e.g. importing the resulting table into QIIME2 or a disconnect of fragment sequences between the table and SEPP generated phylogenies. It also posed the risk of distributing read count to two different versions of the same sequence, when one version was all uppercase, the other with one or more lower case characters. With this bugfix-release: Deblur count tables only contain features with all uppercase nucleotide sequences.

Official version 1.0.3 Released on 19 October 2017.

Features

* Added `--left-trim-length` to allow for trimming nucleotides on the 5' end of each sequence. Please see [issue 154](https://github.com/biocore/deblur/issues/154) for more information.

Performance enhancements
* Updated MAFFT version requirement from 7.221 to 7.310.
* Updated vsearch version requirement from 2.0.3 to >= 2.0.3

Bug fixes
E-value thresholding removed in favor of a bitscore threshold which is scaled based on the length of the query sequence. This is to make each query sequence independent of each other on assessment against the positive filtering database. Please see [PR 146](https://github.com/biocore/deblur/pull/146) for further details.

Official version 1.0.1 Released on 28 February 2017.

Features

Backward-incompatible changes [stable]

Performance enhancements

Bug fixes
* Filtering thresholds for the reference database were not in use. What this means is that, in some cases, highly artifactual sequence could filter through. Specifically, no coverage and similarity thresholds were used with SortMeRNA for the postive reference filter. These are now set to 60 and 65 respectively. In addition, the e-value threshold used is now the SortMeRNA default of 1. Please see [issue 139](https://github.com/biocore/deblur/issues/139) for further discussion in addition for the rational behind the parameter values picked.