eiMakeDb: Create an embedded database
In eiR: Accelerated similarity searching of small molecules

Description Usage Arguments Details Value Author(s) See Also Examples

View source: R/core.R

Uses the initalized compound data to create an embedded compound databbase with r reference compounds in d dimensions.

	eiMakeDb(refs,d,descriptorType="ap",distance=getDefaultDist(descriptorType), 
				dir=".",numSamples=getGroupSize(conn,
				name = file.path(dir,Main)) * 0.1,conn=defaultConn(dir),
				cl=makeCluster(1,type="SOCK",outfile=""),connSource=NULL,numTrees=100)

`refs`	The reference compounds to use to build the database you wish to query against. `Refs` can be one of three things. It can be a filename of an iddb file giving the index values of the reference compounds to use, it can be vector of index values, or it can be a scalar value giving the number of randomly selected references to use.
`d`	The number of dimensions used to build the database you wish to query against.
`descriptorType`	The format of the descriptor. Currently supported values are "ap" for atom pair, and "fp" for fingerprint.
`distance`	The distance function to be used to compute the distance between two descriptors. A default function is provided for "ap" and "fp" descriptors.
`dir`	The directory where the "data" directory lives. Defaults to the current directory.
`numSamples`	The number of non-reference samples to be chosen now to be used later by the eiPerformanceTest function.
`conn`	Database connection to use.
`cl`	A SNOW cluster can be given here to run this function in parrallel.
`connSource`	A function returning a new database connection. Note that it is not suffient to return a reference to an existing connection, it must be a distinct, new connection. This is needed for cluster operations that make use of the database as they will each need to craete a new connection. If not given, certain parts of this function will not be parrallelized. This function can also be used to setup the envrionment on the cluster worker nodes. For example, you might need to re-load libraries like RSQLite and such.
`numTrees`	Affects the build time and the index size. A larger value will produce more accurate results, but use more disk space. See https://github.com/spotify/annoy for more details.

This function will embedd compounds from the data directory in another space which allows for more efficient searching. The main two parameters are r and d. r is the number of reference compounds to use and d is the dimension of the embedding space. We have found in practice that setting d to around 100 works well. r should be large enough to “represent” the full compound database. Note that an r by r matrix will be constructed during the course of execution, so r should be less than about 46,000 to avoid overflowing an integer. Since this is the longest running step, a SNOW cluster can be provided to parallelize the task.

To help tune these values, eiMakeDb will pick numSamples non-reference samples which can later be used by the eiPerformanceTest function.

eiMakdDb does its job in a job folder, named after the number of reference compounds and the number of embedding dimensions. For example, using 300 reference compounds to generate a 100-dimensional embedding (r=300, d=100) will result in a job folder called run-300-100. The embedding result is the file matrix.<r>.<d>. In the above example, the output would be run-300-100/matrix.300.100.

Creates files in dir ("run-r-d" by default). The return value is an id number called the runId, which needs to be given to other functions such as eiQuery or eiAdd.

Kevin Horan

eiInit eiPerformanceTest eiQuery eiCluster

   library(snow)

   r<- 50
   d<- 40

   #initialize 
   data(sdfsample)
   dir=file.path(tempdir(),"makedb")
   dir.create(dir)
   eiInit(sdfsample,dir=dir,skipPriorities=TRUE)

   #create compound db
   runId=eiMakeDb(r,d,numSamples=20,dir=dir,
      cl=makeCluster(1,type="SOCK",outfile=""))