loadSdf: Load SDF and SMILES Data

In girke-lab/ChemmineR: Cheminformatics Toolkit for R

Load SDF and SMILES Data

Description

Load an SDF or SMILES formatted file or SDFSet objects into the database. This will also load arbitrary features from the data as well as descriptor data. The fct parameter can be used to specify a function which will compute features which will then be indexed and stored in the database. These features can later be used to quickly search for compounds. Descriptors can also be computed and stored in another table.

Usage

	loadSdf(conn, sdfFile, fct = function(x) data.frame(), descriptors=function(x) data.frame(descriptor=c(),descriptor_type=c()),
				Nlines = 50000, startline = 1, restartNlines = 1e+05,updateByName=FALSE)
	loadSmiles(conn, smileFile, ...)

Arguments

conn	A database connection object, such as is returned by initDb.
sdfFile	Either the filename of an SDF formated file, or and SDFSet object.
smileFile	The filename of an SMILES formated file.
...	When calling loadSmiles, any of the arguments for loadSdf can be used and will be passed to loadSdf internally.
fct	A function to extract features from the data. It will be handed an SDFSet generated from the data being loaded. This may be done in batches, so there is no guarantee that the given SDFSset will contain the whole dataset. This function should return a data frame with a column for each feature and a row for each compound given, and in the same order. Each of these features will become a new, indexed, table in the database which can be used later to search for compounds. The column name will become the feature name. If not given, no features are computed.
descriptors	This function will also be given an SDFSet object, which may be done in batches. It should return a data frame with the following two columns: "descriptor" and "descriptor_type". The "descriptor" column should contain a string representation of the descriptor, and "descriptor_type" is the type of the descriptor. Our convention for atom pair is "ap" and "fp" for finger print. The order should be maintained. If not given no descriptors are computed.
Nlines	When reading data from a file, the number of lines to read at a time. See also sdfStream.
startline	When reading data from a file, the line number to start reading from.See also sdfStream
restartNlines	When reading data from a file and startline > 1, the number of lines to look forward to find the start of the next compound. See also sdfStream
updateByName	If true we make the assumption that all compounds, both in the existing database and the given dataset, have unique names. This function will then avoid re-adding existing, identical compounds, and will update existing compounds with a new definition if a new compound definition with an existing name is given. If false, we allow duplicate compound names to exist in the database, though not duplicate definitions. So identical compounds will not be re-added, but if a new version of an existing compound is added it will not update the existing one, it will add the modified one as a completely new compound with a new compound id.

Details

Arguments to loadSmiles are the same as those to loadSdf. LoadSmiles will convert its input into an SDFSet and then call loadSdf.

New features can also be added using this function. However, all compounds must have all features so if new features are added to a new set of compounds, all existing features must be computable by the fct function given. If new features are detected, all existing compounds will be run through fct in order to compute the new features for them as well.

For example, if dataset X is loaded with features F1 and F2, and then at a later time we load dataset Y with new feature F3, the fct function used to load dataset Y must compute and return features F1, F2, and F3. loadSdf will call fct with both datasets X and Y so that all features are available for all compounds. If any features are missing an error will be raised.

If just new features are being added, but no new compounds, use the addNewFeatures function.

Value

Returns the compound id numbers of each compound loaded. These can be used to retrieve compounds later. These are id numbers computed by the database and are not extracted from the compound data itself.

Author(s)

Kevin Horan

See Also

sdfStream

Examples

	  
   #create and initialize a new SQLite database
   conn = initDb("test6.db")

	data(sdfsample)

	#just load the data with no features or descriptors
	ids=loadSdf(conn,sdfsample)
	unlink("test6.db")

   conn = initDb("test5.db")
	#load data and compute 3 features: molecular weight, with the MW function, 
	# and counts for RINGS and AROMATIC, as computed by rings, which returns a data frame itself.
	ids=loadSdf(conn,sdfsample,
			  function(sdfset) 
					data.frame(MW = MW(sdfset),  rings(sdfset,type="count",upper=6, arom=TRUE))
			 )
	unlink("test5.db")

girke-lab/ChemmineR documentation built on July 28, 2023, 10:36 a.m.