Quantitative genetic interaction scores allow for more precise functional characterization than binary scores do

Copyright information:

Taken from "A strategy for extracting and analyzing large-scale quantitative epistatic interaction data"

Genome Biology 2006;7(7):R63-R63.

Published online 21 Jul 2006


Different methods for scoring genetic interactions were compared by their propensity to yield high correlations between the profiles of pairs of genes that were assigned to the same functional category. For the S scores and for a binary score in which interactions were classified as either synthetic or noninteracting according to a threshold in the S score, gene pairs were sorted from highest correlation to lowest correlation. The curves show the cumulative number of gene pairs belonging to the same functional category versus the total number of gene pairs for the full S score (blue) and for the binary score (green). The red curve indicates the expected result if the gene pairs are sorted randomly. Inset is a bar graph showing integrations of the two curves over the full range of gene pairs after subtracting the background of the random expectation. The threshold for the binary score was chosen to maximize the integral shown in the inset. The minimal cluster including the seven ALG genes and the four OST genes when clusters are made using hierarchal clustering with the correlation between the profiles of S scores used as the metric. The minimal set of clusters containing the ALG and OST genes when the same clustering algorithm is performed using the binary scores rather than the S scores.