January 2007 2.0.1
This release is a broad-sweeping bug fix release with a dramatically improved
introduction in the documentation. It includes many simple and complex bug
fixes including; conceptual clarifications of several algorithms, fixes and
improvements to the data representation, several advances in the pKa model, and
significantly more documentation. While this is only a bug-fix release, it
represents a significant step-forward in the refinement of the product. We
hope you find it suits your needs.
Bug Fixes and Refinements:
- Added three additional filters that are designed around the best-selling
prescription drugs
- Fixed two bugs in the algorithm for recognizing hydrogen-bond donors and acceptors
- Abbott Bioavailability Score corrected to be reported as a probability
- Aggregator function augmented to include more than 250 additional known aggregators
- Predicted aggregator QSAR model introduced to further enhance aggregator recognition capability
- XLogP and LogS corrected to always examine the hypervalent form of molecules (as are used in the training set)
- Improved data reporting and filter control for the pharmacokinetic properties
- Fixed category reporting for solubility calculation
- Changed the default TPSA calculation to ignore rather than include P and S surface area
- Renamed and clarified aliphatic_group filter to ``maximum connected non-ring atoms''
- Fixed typos in names beta_carbonyl_quat_nitrogen and t_butoxycarbonyl_tBOC
- Updated chloramidine filter name to the more common imidoyl_chloride
- Fixed error in beta_azo_carbonyl pattern
- Made patterns for cytochalasin, monensin, squalestatin and saponin more specific
- Clarified the exclusion patterns for acyclic_NCN and SCN2
- Dramatically improved the specificity and range of the michael_acceptor pattern
- Implemented an algorithm for ``total functional group count'' that takes account of all functional groups rather than only those included in the filter patterns
- Fixed error in alkyl_halide pattern that limited it to bromine and iodine
- Fixed multiple patterns that matched their function group multiple times because of symmetry
- Limited enamine definition to exclude exceedingly delocalized instances
- Added -tableFlag parameter to mark all failure values in the table file with an asterix
- Fix t_butyl_ether pattern to avoid matching t_butyl_alcohols
- Removed dependence of rigid bond count upon implicit or explicit hydrogens
- Extended the unique flag to allow values of ``true'', ``false'' or a filename used to pre-load a file of uniques
- Cleaned-up table data for the select parameter
- Corrected problem that MIN_HALIDE_FRACTION and ALLOWED_ELEMENTS could not properly be removed from the filter file
- Fixed bug so NEWRULE lines with tailing white space are not ignored
- Fixed crash bug that occurred when a filter file was specified that was not a valid file
- Added pKa for phenol groups with significant electron withdrawing groups
- Improved pKa value for barbituate-like compounds
- Clarified alpha-acyl carbon deprotonation
- Removed 1,2,3 triazole pKa rule
- Clarified tertiary amine rule
- Improved efficiency for applying pKa rules to multi-conformer molecules
October 2005 2.0.0
This is the first official "2.0" version of filter. It includes many bug-fixes
and feature expansions based on the long beta period. Some of these include
higher level filtering flags such as Lipinski and other similar measures. This
release also includes a pKa model that was missed in the beta release as well
as non-pH normalization capabilities. This version is makes exporting
filtering data to other programs facile with both table and sdtag data export.
New Features:
- Neutral pKa model optionally applied to all molecules
- Support for molecular graph normalization
- Optional salt file can be used to remove salts from database
- Calculated properties can be attached to molecules as SD tag data
- All filtering data can be saved in tab-separated value format
- Filter on fractional halide weight
- Removal of known aggregator molecules
- Veber bioavailability filtering
- Lipinski violation filtering
- Abbott/Martin bioavailability filtering
- Pharmacopia bioavailability filtering
- Solubility class filtering
- Chiral center count filtering
February 2005 2.0b1
This is the first version of Filter based on OEChem. As such, it does not
alter the graph of your molecules. This is a departure from Filter 1.x, that
was based on OELib and applied a pH model to the molecules in an attempt to fix
them in protonation state for pH=7.0.
This version of filter has also eliminated the need for cumbersome external
data files while still allowing user control over the filtering process.
Several new functional groups have been added to the filters and more realistic
filter values have been used for several functional groups.
High-level filters such as ``total number for functional groups'' and ``carbon
to heteroatom ratio'' have been added to prevent unfunctionalized molecules
from slipping through the filters.
New Features:
- Based on OEChem
- Superior Data Integrity
- Valence checking (broken molecule filter)
- Substructure selection (for selection of filtered reagent lists)
- Not dependent on external data files
- Improved filters and filter properties
- >500 molecules/second with SMILES I/O
- Support for additional platforms
- Documentation
- Improved command-line interface including command-line help
