StabMap

StabMap performs single cell mosaic data integration by building a mosaic data topology, traversing the topology for each reference dataset to project and predict data onto a common embedding, and generating a joint low-dimensional embedding that accounts for all available relevant features.

Key Features:

• Mosaic data topology construction: StabMap builds a mosaic data topology from input data provided in a list format containing data matrices with all relevant features.
• Joint low-dimensional embedding: The output is a joint low-dimensional embedding that takes into account all available relevant features across datasets.
• Expression imputation: Expression imputation can be performed using the StabMap embedding and any of the original data matrices for given reference and query cell lists.

Scientific Applications:

• Single cell mosaic data integration: StabMap integrates single cell datasets that do not share all features, enabling analysis of heterogeneous multi-modal or multi-batch data.
• Cross-dataset projection and prediction: StabMap projects and predicts data onto a common embedding by traversing the mosaic data topology for each reference dataset.

Methodology:

StabMap builds a mosaic data topology, then for each reference dataset traverses the topology to project and predict data onto a common embedding.