- fix memory problem of running whole genome through `signalp`, now splits into 20 chunks, should result in < 200 proteins per run - work around for BUSCO/Augustus training problem if augustus species set to generic - fix typo in v0.2.3
0.2.2
- yet another fix for `goatools` changing format. `goatools v0.6.4` is now required, previous versions are not supported - Added `signalp` search to `funanntoate annotate` and `funannotate compare`. Since `signalp` has to be manually installed and configured, it will only be run if the program is detected and won't be listed as a dependency.
0.2.1
- fix another bug due to `goatools` format changes
0.1.9
- bug fix for GO enrichment using `goatools` as the developers changed the input parameters for the enrichment script - `funannotate predict` now uses hints for `augustus` prediction. hints are generated from `BLAT` transcript alignments from the `--transcript_evidence` option and protein hints are generated from `exonerate` alignments from the `--protein_evidence` option. The `weights` file from `evidence modeler` was adjusted slightly to account for better predictions from `augustus`.
0.1.8
- fix bug in `funannotate compare` where script would crash if gene models that had the same base name - added `funannotate check` a little script to tell you if Python modules are up to date
0.1.7
- fix install `setup.sh` script if first home-brew installation it defaults to asking user for DB install path - updated some docs to reflect some installation changes that were not properly documented in last release