Package 'TFMPvalue'

Efficient and accurate P-value computation for Position Weight Matrices

Description

This package provides a novel algorithm that solves the P-value calculation problem given the score based on a Postion Weight Matrices (PWMs), or the reverse problem: finding the score give the desired P-value. This package is an interface to code originally made available by Helene Touzet and Jean-Stephane Varre, 2007, Algorithms Mol Biol:2, 15.

Details

The original code is taken from http://bioinfo.lifl.fr/TFM/TFMpvalue/TFM-Pvalue.tar.gz, retrived 26/03/2014.

The algorithm is described in Touzet, H., and Varre, J.-S. (2007). Efficient and accurate P-value computation for Position Weight Matrices. Algorithms Mol Biol 2, 15.

Author(s)

Ge Tan

---

Compute the score from P-value.

Description

Computes the score threshold associated with P-value p using the algorithm of Beckstette 2006.

Usage

TFMLazyScore(mat, pvalue, bg=c(A=0.25, C=0.25, G=0.25, T=0.25),
             type=c("PFM", "PWM"), granularity=1e-5)

Arguments

mat	The input matrix. It can be a Position Frequency Matrix (PFM) or Position Weight Matrix (PWM) in log ratio. The matrix must have row names with "A", "C", "G", "T".
pvalue	The required P-value.
bg	The background frequency of the sequences. A numeric vector with names "A", "C", "G", "T".
type	The type of input matrix. Can be "PFM" or "PWM".
granularity	The granularity used in the computation.

Value

The score is returned based on the matrix, given P-value and granularity.

Author(s)

Ge Tan

Examples

## This example is not tested due to running time > 5s
  pfm <- matrix(c(3, 5, 4, 2, 7, 0, 3, 4, 9, 1, 1, 3, 3, 6, 4, 1, 11,
                  0, 3, 0, 11, 0, 2, 1, 11, 0, 2, 1, 3, 3, 2, 6, 4, 1,
                  8, 1, 3, 4, 6, 1, 8, 5, 1, 0, 8, 1, 4, 1, 9, 0, 2, 3,
                  9, 5, 0, 0, 11, 0, 3, 0, 2, 7, 0, 5),
                nrow = 4, dimnames = list(c("A","C","G","T"))
                )
  bg <- c(A=0.25, C=0.25, G=0.25, T=0.25)
  pvalue <- 1e-5
  type <- "PFM"
  granularity <- 1e-5
  TFMLazyScore(pfm, pvalue, bg, type, granularity)

---

Compute score from P-value.

Description

Computes the score threshold associated with a P-value.

Usage

TFMpv2sc(mat, pvalue, bg=c(A=0.25, C=0.25, G=0.25, T=0.25),
         type=c("PFM", "PWM"))

Arguments

mat	The input matrix. It can be a Position Frequency Matrix (PFM) or Position Weight Matrix (PWM) in log ratio. The matrix must have row names with "A", "C", "G", "T".
pvalue	The required P-value.
bg	The background frequency of the sequences. A numeric vector with names "A", "C", "G", "T".
type	The type of input matrix. Can be "PFM" or "PWM".

Value

The score is returned based on the matrix, given P-value.

Author(s)

Ge Tan

References

Touzet, H., and Varre, J.-S. (2007). Efficient and accurate P-value computation for Position Weight Matrices. Algorithms Mol Biol 2, 15.

Examples

pfm <- matrix(c(3, 5, 4, 2, 7, 0, 3, 4, 9, 1, 1, 3, 3, 6, 4, 1, 11,
                  0, 3, 0, 11, 0, 2, 1, 11, 0, 2, 1, 3, 3, 2, 6, 4, 1,
                  8, 1, 3, 4, 6, 1, 8, 5, 1, 0, 8, 1, 4, 1, 9, 0, 2, 3,
                  9, 5, 0, 0, 11, 0, 3, 0, 2, 7, 0, 5),
                nrow = 4, dimnames = list(c("A","C","G","T"))
                )
  bg <- c(A=0.25, C=0.25, G=0.25, T=0.25)
  pvalue <- 1e-5
  type <- "PFM"
  score <- TFMpv2sc(pfm, pvalue, bg, type)

---

Compute P-value from score.

Description

Computes the P-value associated with a score threshold.

Usage

TFMsc2pv(mat, score, bg=c(A=0.25, C=0.25, G=0.25, T=0.25),
         type=c("PFM", "PWM"))

Arguments

mat	The input matrix. It can be a Position Frequency Matrix (PFM) or Position Weight Matrix (PWM) in log ratio. The matrix must have row names with "A", "C", "G", "T".
score	The required score.
bg	The background frequency of the sequences. A numeric vector with names "A", "C", "G", "T".
type	The type of input matrix. Can be "PFM" or "PWM".

Value

The P-value is returned based on the matrix, given the desired score.

Author(s)

Ge Tan

References

Touzet, H., and Varre, J.-S. (2007). Efficient and accurate P-value computation for Position Weight Matrices. Algorithms Mol Biol 2, 15.

Examples

pfm <- matrix(c(3, 5, 4, 2, 7, 0, 3, 4, 9, 1, 1, 3, 3, 6, 4, 1, 11,
                  0, 3, 0, 11, 0, 2, 1, 11, 0, 2, 1, 3, 3, 2, 6, 4, 1,
                  8, 1, 3, 4, 6, 1, 8, 5, 1, 0, 8, 1, 4, 1, 9, 0, 2, 3,
                  9, 5, 0, 0, 11, 0, 3, 0, 2, 7, 0, 5),
                nrow = 4, dimnames = list(c("A","C","G","T"))
                )
  bg <- c(A=0.25, C=0.25, G=0.25, T=0.25)
  score <- 8.77
  type <- "PFM"
  pvalue <- TFMsc2pv(pfm, score, bg, type)

---