Skip to main content
. 2020 Apr 14;36(11):3457–3465. doi: 10.1093/bioinformatics/btaa150

Fig. 1.

Fig. 1.

Workflow for gene classification pipeline. Four methods are compared: SL-A, SL-I, LP-A and LP-I. Model performance on a variety of gene classification tasks is evaluated over a number of different molecular networks, validation schemes and evaluation metrics. Additionally, the performance of SL using node embeddings as features (SL-E) is evaluated (not shown in this figure)