The previous section demonstrated reading and writing SMILES strings from standard input. This required the programmer to perform the file I/O explicitly. Whilst this may be reasonable for SMILES strings that may be read via STL getline, it isn’t suitable for more complex file formats. To ease this task, OEChem TK provides the oemolstream abstraction. The classes oemolistream and oemolostream allow input and output of molecules using the OEReadMolecule and OEWriteMolecule functions, respectively.
Listing 1: High-level Molecule I/O using molstreams
from openeye import oechem ifs = oechem.oemolistream() ofs = oechem.oemolostream() mol = oechem.OEGraphMol() while oechem.OEReadMolecule(ifs, mol): oechem.OEWriteMolecule(ofs, mol)
Listing 1 will read molecules from stdin in SMILES format and write them to stdout in absolute SMILES format. Notice that in this example, the following is done automatically for you by OEReadMolecule:
OEReadMolecule will skip any invalid molecules automatically.
OEChem TK provides the following more Python-like generator methods to read molecules out of a oemolistream if the user prefers. It is important to realize that the same OEGraphMol instance is reused for every molecule being read from the oemolistream.
Generator methods for reading moleculesfrom openeye import oechem ifs = oechem.oemolistream() ofs = oechem.oemolostream() for mol in ifs.GetOEGraphMols(): oechem.OEWriteMolecule(ofs, mol)
Since the same OEGraphMol instance is being reused for every molecule in the stream, special precautions must be taken to load up an entire file into memory. Listing 2 will read molecules into memory by making sure there is a copy made when inserting the molecule into the container. This is useful if the molecules need to be repeatedly accessed.
Listing 2: Reading molecules into memory
from openeye import oechem ifs = oechem.oemolistream() mollist =  for mol in ifs.GetOEGraphMols(): mollist.append(oechem.OEGraphMol(mol))
In addition to SMILES strings, OEChem TK is able to read numerous other molecular file formats, including MDL SD files, Tripos Tripos MOL2 files and PDB files. The format of an input file or stream may be associated with an oemolistream using the SetFormat method, and may be retrieved with GetFormat. The file format is represented by an unsigned integer, which should be one of the constants defined in the OEFormat namespace. A value of OEFormat_UNDEFINED means that there is no file format associated with the oemolstream.
Listing 3: Explicitly setting file formats
from openeye import oechem ifs = oechem.oemolistream() ofs = oechem.oemolostream() ifs.SetFormat(oechem.OEFormat_SDF) ofs.SetFormat(oechem.OEFormat_MOL2) for mol in ifs.GetOEGraphMols(): oechem.OEWriteMolecule(ofs, mol)
|OEFormat_CSV||Comma separated values||Yes||Yes|
|OEFormat_FASTA||FASTA protein sequence||Yes||Yes|
|OEFormat_INCHIKEY||IUPAC InChI Key||No||Yes|
|OEFormat_ISM||Canonical isomeric SMILES||Yes||Yes|
|OEFormat_MDL||MDL Mol File||Yes||Yes|
|OEFormat_MF||Molecular Formula (Hill order)||No||Yes|
|OEFormat_MOL2||Tripos Sybyl mol2 file||Yes||Yes|
|OEFormat_MOL2H||Sybyl mol2 with explicit hydrogens||Yes||Yes|
|OEFormat_MOPAC||MOPAC file format(s)||No||Yes|
|OEFormat_PDB||Protein Databank PDB file||Yes||Yes|
|OEFormat_RDF||MDL RDF File||Yes||No|
|OEFormat_SDF||MDL SD File||Yes||Yes|
|OEFormat_SKC||MDL ISIS Sketch File||Yes||No|
|OEFormat_SLN||Tripos Sybyl Line Notation||No||Yes|
|OEFormat_SMI||Canonical isomeric SMILES||Yes||Yes|
|OEFormat_XYZ||XMol XYZ format||Yes||Yes|
|OEFormat_OEZ||Zstd Compressed OEBinary||Yes||Yes|
In addition to standard in and standard out, OEChem TK‘s oemolstreams also support reading from files. To open a file, use the open method with the required filename. For input oemolistreams, if the file does not exist the open fails and returns false, for output oemolostreams, the output file is created if it did not previously exist and overwritten if it did. If a filename is passed to the open method, the format and gz compression flags are set based on the file extension of the filename for the oemolistream or oemolostream instance. If no arguments are passed to open, an oemolistream will use standard in, and an oemolostream will use standard out, the format will be set to OEFormat_Default and gz compression will be disabled. Much like regular file I/O, oemolstreams can be closed after use, using the close method. oemolstreams are automatically closed by their destructors.
Listing 4: Reading and writing molecule files
from openeye import oechem ifs = oechem.oemolistream() ofs = oechem.oemolostream() if ifs.open("input.sdf"): if ofs.open("output.mol2"): for mol in ifs.GetOEGraphMols(): oechem.OEWriteMolecule(ofs, mol) else: oechem.OEThrow.Fatal("Unable to create 'output.mol2'") else: oechem.OEThrow.Fatal("Unable to open 'input.sdf'")
One convenient feature of the open method of oemolstreams is that it sets the file format and gz compression associated with the stream from the file extension of the filename used as an argument. Listing 4 converts the file ‘input.sdf’ in MDL file format into the file ‘output.mol2’ in Tripos MOL2 file format. This behavior can be overridden by calling SetFormat after the open but before the first molecule is read or written to a stream.
The oemolistream and oemolostream provide the ability to read and write memory buffers instead of files. This is accomplished through the openstring methods, oemolistream.openstring and oemolostream.openstring.
For any of the molecular file formats supported by OEChem TK, it is often convenient to read and write compressed files or strings. Molecule streams support gzipped input and output via the zlib library. The .gz suffix on any filename used to open a stream is recognized and the stream is read or written in compressed format. This mechanism does not interfere with the format perception. For instance, foo.sdf.gz is recognized as a gzipped file in MDL’s SD format.
The Listing 5 example below converts all of the molecules in a gzipped SD format file into a gzipped OEBinary file.
Listing 5: Reading and writing compressed molecule files
from openeye import oechem ifs = oechem.oemolistream("input.sdf.gz") ofs = oechem.oemolostream("output.oeb.gz") for mol in ifs.GetOEGraphMols(): oechem.OEWriteMolecule(ofs, mol)
The OEChem TK‘s binary file format OEFormat_OEB is well suited (and our preferred way) to store multi-conformer molecules. When working with multi-conformer molecule of +1M datasets, the generated files can easily reach a size when transferring them between computers or networks can be slow. One way to handle this issue is to compress the files with gzip before transfer. As the above example shows OEChem TK seamlessly handles gzipped molecule files with gz extension. Alternatively, OEChem TK also provides a native compressed file format, called OEFormat_OEZ, with the oez file extension. This file format currently only supports multi-conformer molecules derived from the OEMCMolBase class.
One of the main differences between the gzipped OEFormat_OEB and the OEFormat_OEZ files that while gzip compresses a molecular file as a whole, in an OEFormat_OEZ file, each molecule is individually compressed using the Zstandard compression algorithm.
This difference makes OEFormat_OEZ suitable to work really well with OEMolDatabase that is designed to provide fast read-only random access to molecular files. When an OEMolDatabase opens a gzipped molecule file, uncompressing the whole file is necessary to access any molecule in the dataset. When working with a OEFormat_OEZ file, accessing a a specific molecule in the file only triggers the un-compression of the associated data chunk.
The table below lists the file sizes of the same dataset of 1M molecules in uncompressed and compressed formats with different number of conformations per molecule.
|1M molecules with 10 conformations||5.4GB||1.9GB||2.2GB|
|1M molecules with 50 conformations||22GB||5.7GB||5.5GB|
|1M molecules with 200 conformations||69GB||16GB||13GB|
The OEFormat_OEZ file format was introduced in the 2019.Apr toolkits. OpenEye applications or toolkits released before Apr/2019 will not be able to read or write this new file format.
Using the methods outlined above, it is possible to allow the stream format to be controlled from the command line. OEChem TK‘s oemolstreams control the format by interpreting the input and output file names.
Listing 6: Controlling File Format from the Command Line
from openeye import oechem import sys if len(sys.argv) != 3: oechem.OEThrow.Usage("%s <input> <output>" % sys.argv) ifs = oechem.oemolistream() ofs = oechem.oemolostream() if not ifs.open(sys.argv): oechem.OEThrow.Fatal("Unable to open %s" % sys.argv) if not ofs.open(sys.argv): oechem.OEThrow.Fatal("Unable to create %s" % sys.argv) for mol in ifs.GetOEGraphMols(): oechem.OEWriteMolecule(ofs, mol)
prompt> convert file1.sdf file1.smi
This command will convert file1.sdf from SD format to SMILES format. A first extension of this idea allows access to standard in and standard out via the “-” filename. For instance:
prompt> convert file2.mol2 -
This command will convert file2.mol2 in MOL2 format and write the molecules to standard out in SMILES, the default format.
Thus if you have another program, GetFromDatabase, which retrieves molecules from a database and writes them in SMILES format, you can chain it with any OEChem TK program. Using your operating systems redirection commands (e.g. - Unix pipe “|” or redirect “>”) you can move molecules directly from GetFromDatabase to convert without a temporary file:
prompt> GetFromDatabase | convert - file3.sdf
This command will take the SMILES format output from GetFromDatabase and generate an SD format file.
However, to make this concept of using standard in and standard out for piping data really useful, one needs to be able to control the format of standard in and standard out similarly to the way it would be controlled for temporary files. To facilitate this, oemolstreams interpret filenames which are ONLY format extensions to indicate format control for standard in and standard out.
Listing 7: Controlling standard in and standard out File Format
from openeye import oechem ifs = oechem.oemolistream(".sdf") ofs = oechem.oemolostream(".mol2") for mol in ifs.GetOEGraphMols(): oechem.OEWriteMolecule(ofs, mol)
Listing 7 is an example of opening an oemolistream to standard in with the format set to SDF. The output oemolostream is opened to standard out with the format set to MOL2. This is exactly equivalent to Listing 3. However, it demonstrates the extensibility of controlling the format of standard in and standard out from the command line.
This prevents you from naming files as pure extensions: .mol2, .sdf, etc.
Now, using our program convert from Listing 6 it is possible to do the following:
prompt> convert .sdf .mol2
This command opens standard in with the SDF format and opens standard out with the MOL2 format.
Now there is complete format control of standard in and standard out from the command line. If we have a program GenerateStructures, which only writes the MOL2 format and another program GenerateData, which only reads the SD format, we can use them from the command line with any OEChem TK program which uses command-line arguments for file specification:
prompt> GenerateStructures | convert .mol2 .sdf | GenerateData
This command demonstrates how any OEChem TK program with command-line file specification can be used to pipe formatted input and output.
The general goal of the oemolstream input and output classes in OEChem TK is to provide the user with transparent access to the very complex task of reading and writing molecules in a wide variety of formats. However, occasionally, a programmer may want to tweak the behavior of specific readers or writers without abandoning the oemolstreams to use the low level writers (such as OEWriteMDLFile). For these instances, oemolstreams provide the oemolstreambase.SetFlavor and oemolstreambase.GetFlavor methods.
The oemolstreambase.SetFlavor method takes two unsigned integer arguments, the first is the format for which the flavor is being specified and the second is the flavor itself. The formats are specified as discussed in File Formats. The input flavors are specified in the OEIFlavor namespace and the output flavors are specified in the OEOFlavor namespace. Unlike the formats, the flavors are a bitmask and may be binary OR’d together. Under the OEIFlavor and OEOFlavor namespaces, there is a namespace for each format as well as a OEIFlavor_Generic namespace. The OEOFlavor_Generic namespace is used to control aromaticity perception and other properties common to all of formats. To completely specify a flavor, one would typically binary-OR a OEOFlavor_Generic flag and a format specific flag and pass the resultant value to oemolstreambase.SetFlavor.
The default behavior for the PDB reader is that TER specifies the termination of a disconnected fragment within the same molecule while END specifies the termination of a connection table. However, some users may want to have the reader split PDB input files into different molecules every time a TER appears.
The following code is an example of changing the PDB reader flavor.
Listing 8: Changing oemolistream Reader Flavor
from openeye import oechem ifs = oechem.oemolistream('input.pdb') ofs = oechem.oemolostream('output.mol2') flavor = oechem.OEIFlavor_Generic_Default | oechem.OEIFlavor_PDB_Default | oechem.OEIFlavor_PDB_TER ifs.SetFlavor(oechem.OEFormat_PDB, flavor) for mol in ifs.GetOEGraphMols(): oechem.OEWriteMolecule(ofs, mol)
Similar low-level control can be exerted over both input and output oemolstreams using the powerful oemolstreambase.SetFlavor method. See the API documentation for the associated low-level routine and namespace for details on the effects of specific flavor flags.
|File Format||Low-Level Reader||Input Flavors|
|OEFormat_OEZ||N/A (OEReadMolecule works)|
|File Format||Low-Level Writer||Output Flavors|
|OEFormat_OEB||N/A (OEWriteMolecule works)|
|OEFormat_OEZ||N/A (OEWriteMolecule works)|
Low-level readers will not call Clear on the molecule before reading the molecule into it. This is useful for concatenating multiple molecules together.
OEWriteMolecule will change your molecule!
The high-level OEWriteMolecule function standardizes the molecule according to the output type for uniformity. For writing molecules without changing them, there are two options. If you would like the data to appear in the file exactly as it is in the molecule (perhaps Tripos atom names in a .pdb format), then you should use a low level writer. On the other hand, if you would like to write a standardized molecule (e.g. Tripos atom types in a MOL2 file), but don’t want your molecule changed, you can use OEWriteConstMolecule. The OEWriteConstMolecule function makes a copy of the input molecule before performing the necessary perceptions and standardizations.