Skip to main content
Fig. 2 | BMC Bioinformatics

Fig. 2

From: pulver: an R package for parallel ultra-rapid p-value computation for linear regression interaction terms

Fig. 2

Comparison of different input types handled by the R tools lm, MatrixEQTL, and pulver for computation of the linear regression with interaction term. By the braces the dimensions of the matrices are depicted. The R’s build-in function lm can only compute the linear regression with interaction term using one variable with n observations per call. The R package MatrixEQTL is able to compute simultaneously the linear regression for each of p 1 variables from the outcome matrix Y and the interaction term of the matrix X with p 2 variables and the vector Z. In contrast, pulver in addition iterates through p 3variables of the matrix Z and finally computes the linear regression for each column of matrices Y , X and Z

Back to article page