Notable changes

• Now calls indels properly
• Significantly improved variant filter
• Can now be run on samples that do not use barcoded adapters with the --no_barcodes option
  • Identifies duplicates using the first 15 bases of each read
• Now generates summary plots for Train.py and Collapse.py
• Numerous bugfixes

Summary of Changes, but there are far too many to list here:

The wiki has been re-written to reflect these changes. See http://produse.readthedocs.io/en/latest/

General:

• Significant performance improvements for all scripts (in some cases, multiple orders of magnitude)
• "run_produse/ProDuSePipeline.py" can now be parallelized by sample using -j/--jobs
• Now uses a Random Forest classifier to filter variants
  - This is still very much a work in progress
• Removed Stitcher dependency completely, now uses "clip/ClipOverlap.py"
• Framework for INDEL realignment/calling is finished
  - This feature is currently disabled to resolve ongoing bugs
• Reworked config files parsing completely.
  - use "update_config/UpdateConfig.py" to use an older config file with this version of ProDuSe
• sample config file paths can now be specified inside the main config file
• Merged configure_produse.py and bwa.py directly into ProDuSePipeline.py, and removed those scripts
• Now only works with python 3.4 and above
  -local realignment is performed using skbio, which only supports Python 3+

Other changes

• Added Train.py
  - Can be used to train a custom variant classifier
• Barcode sequences are now stored in a FASTQ commend, as opposed to the read name
• Added resume_produse/ResumePipeline.py
  - If a previous instance of ProDuSePipeline.py was terminated, this can be used to restart it
• Tweaks to parameter parsing for all pipeline components
  - Arguments have been renamed to better reflect their function
• Re-wrote the wiki

.... and many more. See commit logs for more details

Bug-fixes and Quality of life improvements

Added documentation (see readme for a link)
Now saves and prints out BWA stderr output in the event of a bwa crash
VAFs are now calculated for all candidate variants (sample.variants.vcf)
Added a version command (produse version)
Fixed bugs when supplying a target region file
Clarified help messages

Re-wrote merged-read splitter: Now no longer causes misalignment of reverse read when splitting
Splitmerge is now callable from the command line directly
Re-wrote filters: Now filters dynamically based upon capture space and locus characteristics
Updated produse_config file to support filter_produse.py
All stages of the ProDuSe pipeline now output status messages

Replaced FLASH with Stitcher
Modified Filters
Significant Usability increases

Version used for non-Hodgkin lymphoma ctDNA evolution work