Commits on Mar 30, 2017

1. Cache AdjList computation - useful for when we fire lots of SMARTS at… …

   … one molecule. In future this should be handed off to the user to manage, this will be easier when AtomRef is introduced.

   

   johnmay committed Mar 30, 2017
   Copy the full SHA

   f04c234 View commit details

   Browse the repository at this point in the history

2. SMARTS mass matching semantics, 0 = unspecified. Attempting to update… …

   … the grammar was not successful.

   

   johnmay committed Mar 30, 2017
   Copy the full SHA

   3231066 View commit details

   Browse the repository at this point in the history

3. Minor optimisation we test either the ring connectivity or number or … …

   …rings. In the case of R0 we can also check connectivity. [!R0] = [R]

   

   johnmay committed Mar 30, 2017
   Copy the full SHA

   6115fed View commit details

   Browse the repository at this point in the history

4. More elegant handling of non-SMARTS patterns with valid SMARTS [!*] m… …

   …atching nothing.

   

   johnmay committed Mar 30, 2017
   Copy the full SHA

   cb87f14 View commit details

   Browse the repository at this point in the history

5. Safe SMARTS optimizations - reorder pattern to be more efficient with… …

   …out changing the meaning.

   

   johnmay committed Mar 30, 2017
   Copy the full SHA

   0e5f60c View commit details

   Browse the repository at this point in the history

6. Use VF matcher.

   

   johnmay committed Mar 30, 2017
   Copy the full SHA

   f6b74f9 View commit details

   Browse the repository at this point in the history

7. Basic non-compatible SMARTS changes. We implement bit1 and allow bit3… …

   … and bi18 to also match the aromatic atoms.

   

   johnmay committed Mar 30, 2017
   Copy the full SHA

   65170a9 View commit details

   Browse the repository at this point in the history

8. Implement ring bits in code instead of SMARTS.

   

   johnmay committed Mar 30, 2017
   Copy the full SHA

   1875025 View commit details

   Browse the repository at this point in the history

9. Count aromatic rings at the same time. When MDL say aromatic they rea… …

   …lly only mean benzene/pyridine - using the correct arom model this change has no change of results.

   

   johnmay committed Mar 30, 2017
   Copy the full SHA

   adb075d View commit details

   Browse the repository at this point in the history

10. Improved fragment detection and compatibility (valid SMARTS).

    

    johnmay committed Mar 30, 2017
    Copy the full SHA

    44aefd4 View commit details

    Browse the repository at this point in the history

11. Remove check for rings >= 8M bit101. bit101 is described as 8M ring b… …

    …ut was actually matching >= 8. Removing this and only checking for 8M rings affects a large number of molecules.

    

    johnmay committed Mar 30, 2017
    Copy the full SHA

    a5331be View commit details

    Browse the repository at this point in the history

12. Replace recursive patterns, the alternatives provided by the recursiv… …

    …e patterns should usually be matched, e.g. bit118: ACH2CH2A is not equivalent to *1~[#6H2]~[#6H2]~1. This is confirmed by other MACCS implementations (T.J. O'Donnell, Design and use of relational databases in chemistry). There are a lot of changes in MACCS FP generate because recursive SMARTS were incorrectly used: *~[#6H2]~[#6H2]~* matches the SMILES OCCO once uniquely, whilst [$(*~[#6H2]~[#6H2]~*)] matches it twice.

    

    johnmay committed Mar 30, 2017
    Copy the full SHA

    0c68f92 View commit details

    Browse the repository at this point in the history

13. FIxed FP size.

    

    johnmay committed Mar 30, 2017
    Copy the full SHA

    f45fb6c View commit details

    Browse the repository at this point in the history