Release 1.6.0

Notable new features and improvements

• Restructure base classes to generalize for oligos [#57]
• Allow primer design with thermodynamic models [#66, #67, #68]
• Change default objective to maximize activity [#70]

Minor changes

• Clean up and improve descriptions of several arguments [9e4b379, c59807c, 8a6c6c3, 484d78f]
• Change continuous integration testing to use GitHub Actions [#71]

Bug fixes

• Fix bug when hashing k-mers, e.g., for representative potential guides [#69]

Release 1.5.0

Notable new features and improvements

• Allow sequences to be weighted, both automatically by subtaxa and manually by a specified file [#63]
• Allow per sequence analysis in analyze_coverage, and add options for a maximum target length and primer terminal mismatches [#59]

Minor changes

• Modify expected activity per guide to only output the expected activity of the sequences for which the guide is the best option [#60]
• Check for the reverse complement of the sequence when aligning with MAFFT [#62]
• Consider consensus of all sequences when selecting guides [#65]

Bug fixes

• Make sure reference accessions are for the correct segment [#61]

Release 1.4.1

Notable new features and improvements

• Add ability to automatically align FASTA files [#58]

Bug fixes

• Allow spaces in metadata filters

Release 1.4.0

Notable new features and improvements

• Add option to design naively using n most common sequences [#54]
• Add genomic annotations to output [#53]

Minor changes

• Redirect taxonomic IDs using NCBI Taxonomy DB [#56]
• Improve performance of coverage analysis against a large number of target sequences

Release 1.3.0

Notable new features and improvements

• Incorporate the likelihood of decaying activity over time (set via --predict-activity-degradation, --predict-activity-degradation-mu, --predict-activity-degradation-t, and --predict-activity-degradation-n ) [#47]
• Add option to specify type of overlap allowable between assay designs via --do-not-overlap [#48]

Minor changes

• Allow using our Cas13a model via --predict-cas13a-activity-model [#52]
• Set the logging level of there being no guides in a given window to the debug level [#51]
• Update TensorFlow to 2.3.2 [#44]

Bug fixes

• Make README examples work when downloading from a package manager [#52]

Release 1.2.0

Notable new features and improvements

• Automatically determines reference sequences from NCBI by default; reference accessions are no longer a positional argument, but can be manually set via --ref-accs [#32]
• Allows filtering taxa based on metadata from NCBI, letting designs be specific to subsets of species and/or specific against subsets of species [#32]

Minor changes

• Adds option to write mean activity of guides in coverage analysis [#43]
• Expands testing: integration/functional testing for design.py [#36] and automated multiOS testing on TravisCI [#42]
• Sets up Bioconda and PyPi packaging [#41, #42]

Bug fixes

None

Release 1.1.1

Notable new features and improvements

• Updates documentation in the README to more clearly describe ADAPT and explain how to run it [#35]
• Adds Dockerfiles to allow containerization for ADAPT images [#31]

Minor changes

• Adds ability to use U instead of T in input sequence FASTAs [#34]
• Adds option to not memoize guide level computations [#33]

Bug fixes

None

Release 1.1.0

Notable new features and improvements

• Adds feature to save memoization data to an AWS S3 Bucket [#29]
• Adds feature to design guides naively based on conservation metrics (specifically, using Shannon entropy) [#30]

Minor changes

• Adds option to set a random seed for all of ADAPT in order to get more reproducible results (Some stochasticity still exists due to object-level hashing) [#29]
• Adds option to design_naively.py to find the best n guides [#30]

Bug fixes

• Fixes bug when writing target stats without predict activity model by using nan for the expected activity [518aafa]

Release 1.0.0

First official release of ADAPT!