Skip to main content
Fig. 2 | Genome Medicine

Fig. 2

From: Exploring the cellular basis of human disease through a large-scale mapping of deleterious genes to cell types

Fig. 2

Overview of the GSC, text-mining and GSO methods. (A) For the GSC method, percentile-normalized relative gene expression scores are integrated with PPI data to create interactomes. Permuted interactomes are created by permuting the expression scores. The compactness score of the disease-associated gene set is computed for each observed and permuted interactome and empirical P values produced by counting the proportion of permuted compactness scores less than the observed compactness score. (B) To complete the text-mining, diseases from DisGeNET and cell types from the FANTOM5 project were mapped to MeSH terms using a number of controlled vocabularies. These MeSH terms were then used to query PubMed and count the number of articles individually and co-mentioning terms. Fisher’s exact test was used to determine whether the number of co-mentioning articles is greater than expected by chance. (C) For the GSO method, percentile-normalized gene expression scores are used to create observed and permuted expression profiles. The mean expression score of the disease-associated gene set is then computed for the observed and permuted expression profiles. Empirical P values are computed by counting the numbers of permuted scores greater that each observed score

Back to article page