===2101===
- initial version
===2102===
- patterns: added >30 SMARTS, replacing unsuitable wildcard SMILES
- biosyn: added 24 SMARTS
- classify: add option to load superclass dataset; does nothing atm
- add
class-subclass.py
for updating superclass dataset
===2103===
- unified dataset handling (all datasets are loaded from given directory)
- data update
===2104===
- classify: remove redundant hits
- classify: get nearest parent GO process if none on class
- classify: add option to get redundant hits as before
- superclass data: more class types, bigger ontology, dead-end diagnostic
===2105===
- all data: monoterpenoids
- test data: monoterpenoids, 14/197 FAIL
- classify: test capability
- multiple SMARTS (denoting sets of patterns) now handled
===2106===
- all data: sesquiterpenoids
- test data: sesquiterpenoids, 7/240 FAIL
===2107===
- all data: diterpenoids
- test data: diterpenoids, 0/177 FAIL
- don't load SMILES/SMARTS if InChI present, 10x speed up of classification
===2108===
- remove bogus pattern leading to Boost warnings
- revised/added to acyclic monoterpenoids, FAIL: 0/197
- all data: sesterterpenoids
- test data: sesterterpenoids, 1/14 FAIL
- split pattern query, makes it faster, more resilient, one third reduction in pattern db size
===2109===
- some new files were missing in last release
- reduce test run output
- all data: triterpenoids
- test data: triterpenoids, 1/85 FAIL
===2110===
- pattern fixes (not all possible double bonds were allowed)
- all data: tetraterpenoids
- test data: tetraterpenoids, 7/199 FAIL
===2111===
- all data: steroids
- test data: steroids, 3/1561 FAIL
===2112===
- add -n option to classify small sample of WD natural products
- all data: flavonoids pt. 1 (est. 60% of all)
- test data: flavonoids
===2113===
- all data: flavonoids pt. 2 (est. 85% of all)
===2114===
- add -j option for JSON output (NP only)
- all data: flavonoids pt. 3 (est. 95% of all)
==2115==
- all data: alkaloids pt. 1 (est. 25% of all)
- test data: alkaloids
==2116==
- fixed bug where compounds with biosyn process were not seen with -j option
- all data: alkaloids pt. 2 (est. 45% of all)
==2117==
- sort pattern data on SMARTS hash to avoid most reshuffling
- all data: alkaloids pt. 3 (est. 62% of all)
==2118==
- fix bug where hits to NP root classes were ignored
- again fix data write order; Py hash() was salted
- add rule to recognize unspecified alkaloids
- all data: alkaloids pt. 4 (est. 35% unspecified)
- add option ouputting NP roots in JSON format
- revamp top NP hierarchy
==2119==
- all data: polyketides pt. 1 (est. 90% of all)
- test data: polyketides
- prettify
--list_nproots
- further NP root optimizations
==2120==
- test data polyketides was missing
- add rule to recognize unspecified macrolides
- all data: polyketides pt. 2 (est. 97% of all)
==2121==
- classify: revamp -n code
- classify: include mol data with JSON output
- improve macrolide rule
- reduced unspec. alkaloids to 25% by adding more small groups
- unspec. alkaloids and makrolides tagged in WD
==2122==
- all data: fatty acyl pt. 1 (est. 77% of all)
- test data: fatty acyl
- classify: fatty acyl rules