mechanochat.tools.contact_lr_communication#

mechanochat.tools.contact_lr_communication(df_ligrec: DataFrame, sem: SEM2 | SEM3 | None = None, adata: AnnData | None = None, contact_key: str | None = 'contacts', lr_delimiter: str = '-', heteromeric_delimiter: str = '_', prefix: str = 'contact')[source]#

Contact-based ligand-receptor communication inference.

For every pair of contacting cells, the signal of a ligand-receptor pair is the product of the ligand expression in the sender and the receptor expression in the receiver, evaluated only on the cell-cell contact graph.

Parameters:

  • df_ligrec (pandas.DataFrame) – Dataframe of ligand-receptor pairs. Required columns: Ligand, Receptor, Pathway.

  • sem (SEM2, SEM3, optional) –

    SEM object providing .contact_matrix (cell-cell contact matrix) and .adata.

    If sem.contact_matrix has not been computed, call sem.compute_contact().

    If None, the contact matrix is read from adata.obsp[contact_key].

    If both sem.adata and adata are provided, adata parameter takes precedence.

  • adata (AnnData, optional) –

    AnnData object that the results are written into if sem is not provided

    If sem is None, the contact matrix is read from adata.obsp[contact_key].

  • contact_key (str, default 'contacts') – Key of the contact matrix in adata.obsp, used only when sem is None.

  • lr_delimiter (str, default '-') – Delimiter used to construct ligand-receptor pair names in output.

  • heteromeric_delimiter (str, default '_') –

    Delimiter used to separate subunits in heteromeric complexes within df_ligrec.

    For example, if a receptor complex is ‘TGFBR1_TGFBR2’, this parameter should be ‘_’.

  • prefix (str, default 'contact') – Prefix of the keys written to .obsp.

Returns:

  • None

  • Sets the following fields in adata –

    • .obsp[‘<prefix>_<ligand>-<receptor>’] : (nc, nc) csr_matrix of cell-cell communication mediated by <ligand>-<receptor>

    • .obsp[‘<prefix>_<pathway>’] : sum over the LR pairs of one pathway

    • .obsp[‘<prefix>_total’] : sum over all pathways

    • .obsm[‘sender_<prefix>_signal’] : (nc, n_signal) DataFrame of row sums, columns prefixed ‘s-’

    • .obsm[‘receiver_<prefix>_signal’] : (nc, n_signal) DataFrame of column sums, columns prefixed ‘r-’

    • .uns[‘<prefix>_signal_info’] : dict with keys ‘signal’, ‘pathway’, ‘total’, ‘db’

Notes

Orientation: s[i, j] is the signal from sender i to receiver j (row = sender, column = receiver).