imputeHD: Impute missing row datasets with multiple hot deck.

In RNAseqNet: Log-Linear Poisson Graphical Model with Hot-Deck Multiple Imputation

Description

imputeHD performs multiple hot-deck imputation on an input data frame with missing rows. Each missing row is imputed with a unique donor. This method requires an auxiliary dataset to compute similaritities between individuals and create the pool of donors.

Usage

imputeHD(X, Y, sigma, m = 50, seed = NULL)

Arguments

X	n x p numeric matrix containing RNA-seq expression with missing rows (numeric matrix or data frame)
Y	auxiliary dataset (n' x q numeric matrix or data frame)
sigma	threshold for hot-deck imputation (numeric, positive)
m	number of replicates in multiple imputation (integer). Default to 50
seed	single value, interpreted as an in integer, used to initialize the random number generation state. Default to NULL (not used in this case)

Details

Missing values are identified by matching rownames in X and Y. If rownames are not provided the missing rows in X are supposed to correspond to the last rows of Y.

Value

S3 object of class HDImputed: a list consisting of

• donors a list. Each element of this list contains the donor pool for every missing observations

• draws a data frame which indicates which donor was chosen for each missing samples

• data a list of m imputed datasets

Author(s)

Alyssa Imbert, [e-mail]

Nathalie Vialaneix, [e-mail]

References

Imbert, A., Valsesia, A., Le Gall, C., Armenise, C., Lefebvre, G. Gourraud, P.A., Viguerie, N. and Villa-Vialaneix, N. (2018) Multiple hot-deck imputation for network inference from RNA sequencing data. Bioinformatics. doi: 10.1093/bioinformatics/btx819.

See Also

chooseSigma, imputedGLMnetwork

Examples

data(lung)
data(thyroid)
nobs <- nrow(lung)
miss_ind <- sample(1:nobs, round(0.2 * nobs), replace = FALSE)
lung[miss_ind, ] <- NA
lung <- na.omit(lung)
imputed_lung <- imputeHD(lung, thyroid, sigma = 2)

RNAseqNet documentation built on July 2, 2020, 4:15 a.m.