A common task in Cheminformatics is the identification of unique or duplicate structures. The unique smiles is commonly used for this.
unique_molecules reads a stream of molecules, and if any have been
seen before, that molecule is discarded.
The usage message is
-l strip to largest fragment
-a compare as tautomers - skeleton and Hcount
-c exclude chiral info - optical isomers will be duplicates
-z exclude cis/trans bonding information
-I ignore isotopic labels
-y all non-zero isotopic values considered equivalent
-f function as filter (TDT input)
-p <fname> specify previously collected molecules
-G <tag> identifier tag when working as a filter
-s <size> specify primary hash size (default 1000)
-S <name> specify output file name stem
-D <name> write duplicate structures to <name>
-R <rxn> perform reaction(s) on molecules before comparing
-T discard molecular changes after comparison
-r <number> report progress every <number> molecules
-e report all molecules together with counts
-j items are the same only if both structure and name match
-t E1=E2 element transformations, enter '-t help' for details
-i <type> specify input type
-o <type> specify output type(s)
-A <qualifier> Aromaticity, enter "-A help" for options
-K ... standard smiles options, enter '-K help' for info
-g <qualifier> chemical standardisations, enter "-g help" for usage
-v verbose output
There are a great many ways by which two molecules can be considered identical.
unique_molecules provides a wide variaty of structural modifiers that can
be applied to molecules before their unique smiles are generated.
- -l strip to the largest fragment
- -c discard chirality
- -z discard cis/trans bonding information
- -I discard isotopic labels.
- -y all non zero isotopic labels compared as equivalent
- -R <rxn> apply a transformation reaction to the molecules.
- -t E1=E2 transform all elements E1 to E2 (suggest -T I=Cl -T Br=Cl)
Once these transformations are applied, the unique smiles is generate and compared to what has been encountered previously. If this is the first instance of the smiles, that molecule is written to the output strea, otherwise it is classified as a duplicate, and if specified, written to the -D file for duplicates.
By default, the changed molecule will be written, but the original form can be written if the -T option is used. That should probably be the default.
If there is a previous set of molecules (-p) that should be considered,
which will establish the unique smiles hash, without writing anything.
By default, smiles are accumulated in a C++ set, which has only keys. If you wish to get a summary of smiles and counts, use the -e option, which will use a map for the unique smiles, and counts can be accumulated.
Some of the other options modify how this works.
Specify a file of previously selected molecules. The unique smiles hash is first populated with these molecles, and then the input is compared.
Obscure. Identifier tag when working as a TDT filter - GFP files.
Ignore. Specify primary hash size (default 1000). Usually not needed these days.
Specify output file name stem.
Write duplicate structures to <name>. By default, duplicate structures are just discarded.
Discard molecular changes after comparison. This is important. For example if chirality is discarded during unique smiles comparison, by default, there will be no chirality in the ouput. With the -T option, the original molecule is retained. This should be the default.
Report progress every <number> molecules processed. Or just look at what is getting written.
It can be helpful to get a summary of unique smiles and counts written to the log.
Formalise what the -e option does, writing the data to a
separate file - rather than mixed in with stderr.
Only consider structures the same if their names also match.