Fig. 1

Visual depiction of the SKiM algorithm. An A term is provided by the user (for example, a disease such as “Raynaud’s”). A list of B-terms (e.g., phenotypes) are each paired with the A–term, and a Fisher’s Exact Test is performed to get a p-value. The significant A–B pairs are ranked and filtered by p-value and are each paired with a list of C terms (e.g., drugs). B–C pairs are sorted by prediction score. The user can provide their own B and C terms or can use pregenerated lists stored on the SKiM website