SnpSet-methods: Accessors and methods for SnpSet objects

In jmacdon/oligoClasses: Classes for high-throughput arrays supported by oligo and crlmm

Description

Utility functions for accessing data in SnpSet objects.

Usage

calls(object)
calls(object) <- value
confs(object, transform=TRUE)
confs(object) <- value

Arguments

object	A SnpSet object.
transform	Logical. Whether to transform the integer representation of the confidence score (for memory efficiency) to a probability. See details.
value	A matrix.

Details

calls returns the genotype calls. CRLMM stores genotype calls as integers (1 - AA; 2 - AB; 3 - BB).

confs returns the confidences associated with the genotype calls. The current implementation of CRLMM stores the confidences as integers to save memory on disk by using the transformation:

round(-1000*log2(1-p)),

where 'p' is the posterior probability of the call. confs is a convenience function that transforms the integer representation back to a probability. Note that if the assayData elements of the SnpSet objects are ff_matrix or ffdf, the confs function will return a warning. For such objects, one should first subset the ff object and coerce to a matrix, then apply the above conversion. The function snpCallProbability for the callProbability slot of SnpSet objects. See the examples below.

checkOrder checks whether the object is ordered by chromosome and physical position, evaluating to TRUE or FALSE.

Note

Note that the replacement method for confs<- expects a matrix of probabilities and will automatically convert the probabilities to an integer representation. See details for the conversion.

The accessor snpCallProbability is an accessor for the 'callProbability' element of the assayData. The name can be misleading, however, as the accessor will not return a probability if the call probabilities are represented as integers.

See Also

The helper functions p2i converts probabilities to integers and i2p converts integers to probabilities.

See order and checkOrder.

Examples

	theCalls <- matrix(sample(1:3, 20, rep=TRUE), nc=2)
	p <- matrix(runif(20), nc=2)
	integerRepresentation <- matrix(as.integer(round(-1000*log(1-p))), 10, 2)
	obj <- new("SnpSet2", call=theCalls, callProbability=integerRepresentation)
	calls(obj)
        confs(obj)  ## coerces to probability scale
        int <- Biobase::snpCallProbability(obj) ## not necessarily a probability
        p3 <- i2p(int)  ## to convert back to a probability

jmacdon/oligoClasses documentation built on Aug. 11, 2020, 12:32 a.m.