CHANGELOG.txt

3.2.0
- revised WDLs for simpler configuration with Terra
- using all common variants for training ML in boosting
- drop SPLICEADJ, now just using DJ
- now explicitly writing feature matrices for use with the downstream ML steps for improved transparency and flexibility
- our RVBLR and py-snpir are now available as submodules, although not fully integrated into the ctat_mutations, and also leverage feature matrices as above.
- uses CTAT genome lib: https://data.broadinstitute.org/Trinity/CTAT_RESOURCE_LIB/__genome_libs_StarFv1.10/


3.0.1
* note: 3.0.1 - bugfix, new syntax for CreateSequenceDictionary with gatk 4.1.9.0 in mut lib prep step
    docker and singularity images provided.


3.0.0
- improved boosting performance
- rewrote workflow logic using WDL
- run 'make' in the base directory to install Cromwell workflow runner
- moved RVBLR (RVBoost) to https://github.com/broadinstitute/RVBLR


2.5.0
- leverages Open-CRAVAT for annotations in place of REST-call to web app.
- incorporates additional boosting methods: SGBoost, GBoost, AdaBoost, and RF
- cleaner organization of output files
- added gnomad pop AF annotations
- added clinvar annotations
- igv-reports incorporates clinvar and FATHMM
- cancer-related variants selected based on chasm & vest pvals, or FATHMM or clinvar pathogenic attributes.

2.4.0
- added support for using single-end rna-seq reads
- support for RVBLR (our RVBoost-integration as RVB-like R)
- pctextpos is computed as an annotation and leveraged by RVBLR
- more robust variant annotation / multithreading
- annot_PASS_reads on by default now (again), but there's an option to disable if necessary.
- various bugfixes