As the development of new drugs reaches its physical and financial limits, drug repurposing has become moreimportant than ever. For mechanistically grounded drug repurposing, it is crucial to uncover the diseasemechanisms and to detect clusters of mechanistically related diseases. Various methods for computing candidatedisease mechanisms and disease clusters exist. However, in the absence of ground truth, in silico validation ischallenging
To address this problem, we present DIGEST, a Python-based tool for in silico validation of disease and genesets, clusterings, or subnetworks. DIGEST enables fully automated validation of gene sets based on theirfunctional similarity calculated on shared associated biological functions and processes and diseases based ontheir genetic similarity. The similarities are used as distance measures in clusterings and the score is determinedby metrics for evaluating clustering algorithms, such as the Dunn index. A variety of user input types aresupported, such as gene or disease sets, clusterings, or subnetworks. DIGEST supports all widely used IDannotation types (e.g. MONDO and Uniprot). The functional and genetic similarity of the user input isstatistically evaluated against a random background model to generate empirical p-values. The results can beeasily visualized with multiple figures showing the calculated empirical p-values and the mappability of the inputIDs. Finally, the user also has the option of checking the significance contribution of each individual ID separately,so that outliers in the user input are easier to identify.
Read the full paper here: https://doi.org/10.1093/BIB/BBAC247.