Skip to main content
Fig. 1 | Genome Biology

Fig. 1

From: Integrating temporal single-cell gene expression modalities for trajectory inference and disease prediction

Fig. 1

Schematic overview of benchmarking design. A Workflow of integration method evaluation. Ten integration approaches and four temporal mRNA modalities are evaluated on three prediction tasks. Data are first preprocessed and jointly batch effect corrected. Next cross-modality integration (spliced and unspliced counts or moments of spliced and RNA velocity) is performed using ten different integration approaches. Features specified through the integration strategy are used to infer trajectories, predict response to drug treatment, and classify patient cells. B Overview of data integration strategies (unintegrated, concatenation, sum, CellRank [17], Seurat v4 [49], MOFA+ [30], Grassmann joint embedding [26], integrated diffusion [24], SNF [25], and PRECISE [50])

Back to article page

Genome Biology

ISSN: 1474-760X

Contact us