score.cal {MLDA}R Documentation

Weighted Score Scheme

Description

A function to assign each feature a weighted score based on the cut-offs for methylation and unmethylation.The averaged scores for two calsses are calculated. A robust regresion model is constructed using the averaged scores for two classes. The positive and negative outliers of regression models are identified.

Usage

score.cal(mlda.obj, cutoff.obj, mlda.data, perm = 200, control = NULL)

Arguments

mlda.obj An object from mlda()
cutoff.obj An object from cutoff()
mlda.data A dataset object frmo mlda.data()
perm The times of sampling in the normal distribution estimated from the standerdised residuals of robust regression model construced by averaged sensitive and resistant scores
control The mean and variance of normal distribution are estimated using controls of which methylations status are not changed in two classes

Details

Each feature was scored based on the cut-offs of log likelihood ratios for methylation and unmethylation on dye-swapped/duplicate arrays using a weighted methylation scoring scheme. The features consistently identified as methylated candidates on dye-swapped/duplicate arrays are scored of 1; similarly unmethylated features were scored of -1. The rest of the feature were assigned a weighted score based on their location on the plot (see lr.plot).

Score scheme:

f ( i, j ) = 1 i > a and j > a
f ( i, j ) = -1 i > b and j > b
f ( i, j ) = 0.45 ( i > a and 0 < j < a ) or ( 0 < i < a and j > a )
f ( i, j ) = -0.45 ( i < b and b < j < 0 ) or ( b < i < 0 and j < b )
f ( i, j ) = 0.2 ( 0 < i < a and 0 < j < a ) or ( b < i < 0 and b < j < 0 )
f ( i, j ) = 0.0001 ( 0 < i < a and b < j < 0 ) or ( b < i < 0 and 0 < j < a )

otherwise, f( i, j ) = 0

a: methylation cut-off
b: unmethylation cut-off

Value

A list containing:

sr.obj A list including standardised residuals, p value for positive and negative outliers at various thresholds
score A score matrix. Each feature is assigned a score.

Author(s)

Wei Dai w.dai@imperial.ac.uk

See Also

MLDA Examples are included


[Package MLDA version 1.0-1 Index]