napistu 0.2.5.dev6__py3-none-any.whl → 0.3.1__py3-none-any.whl

napistu/ontologies/dogma.py

@@ -0,0 +1,243 @@
+from __future__ import annotations
+
+import logging
+import pandas as pd
+
+from napistu import sbml_dfs_core
+from napistu import source
+from napistu import identifiers
+from napistu import utils
+from napistu.ontologies.genodexito import Genodexito
+from napistu.constants import BQB
+from napistu.constants import IDENTIFIERS
+from napistu.constants import MINI_SBO_FROM_NAME
+from napistu.constants import ONTOLOGIES
+from napistu.constants import SBML_DFS
+from napistu.ontologies.constants import INTERCONVERTIBLE_GENIC_ONTOLOGIES
+from napistu.ontologies.constants import GENE_ONTOLOGIES  # noqa: F401
+from napistu.ontologies.constants import GENODEXITO_DEFS
+from napistu.ontologies.constants import NAME_ONTOLOGIES
+from napistu.ontologies.constants import PROTEIN_ONTOLOGIES
+
+logger = logging.getLogger(__name__)
+
+
+def create_dogmatic_sbml_dfs(
+    species: str,
+    preferred_method: str = GENODEXITO_DEFS.BIOCONDUCTOR,
+    allow_fallback: bool = True,
+    r_paths: str | None = None,
+) -> sbml_dfs_core.SBML_dfs:
+    """
+    Create Dogmatic SMBL_DFs
+
+    Create an SBML_dfs model which is pretty much just proteins and no
+    reactions, as well as annotations linking proteins to genes, and
+    creating nice labels for genes/proteins.
+
+    Args:
+        species (str):
+            An organismal species (e.g., Homo sapiens)
+        r_paths (str or None)
+            Optional, p]ath to an R packages directory
+
+    Returns:
+        dogmatic_sbml_dfs (sbml.SBML_dfs)
+            A pathway model which (pretty much) just contains proteins and
+            diverse identifiers
+    """
+
+    dogmatic_mappings = _connect_dogmatic_mappings(
+        species, preferred_method, allow_fallback, r_paths
+    )
+
+    logger.info("Creating inputs for sbml_dfs_from_edgelist()")
+
+    # format entries for sbml_dfs_from_edgelist()
+    species_df = dogmatic_mappings["cluster_consensus_identifiers_df"].join(
+        dogmatic_mappings["s_name_series"]
+    )
+
+    # stub required but invariant variables
+    compartments_df = sbml_dfs_core._stub_compartments()
+    interaction_source = source.Source(init=True)
+
+    # interactions table. This is required to create the sbml_dfs but we'll drop the info later
+    interaction_edgelist = species_df.rename(
+        columns={
+            "s_name": "upstream_name",
+            SBML_DFS.S_IDENTIFIERS: SBML_DFS.R_IDENTIFIERS,
+        }
+    )
+    interaction_edgelist["downstream_name"] = interaction_edgelist["upstream_name"]
+    interaction_edgelist["upstream_compartment"] = "cellular_component"
+    interaction_edgelist["downstream_compartment"] = "cellular_component"
+    interaction_edgelist["r_name"] = interaction_edgelist["upstream_name"]
+    interaction_edgelist["sbo_term"] = MINI_SBO_FROM_NAME["reactant"]
+    interaction_edgelist["r_isreversible"] = False
+
+    dogmatic_sbml_dfs = sbml_dfs_core.sbml_dfs_from_edgelist(
+        interaction_edgelist=interaction_edgelist,
+        species_df=species_df,
+        compartments_df=compartments_df,
+        interaction_source=interaction_source,
+        upstream_stoichiometry=-1,
+        downstream_stoichiometry=1,
+        downstream_sbo_name="product",
+    )
+
+    # remove all reactions except 1 (so it still passes sbml_dfs.validate())
+    # this self reaction will be removed when creating the graph
+    dogmatic_sbml_dfs.remove_reactions(dogmatic_sbml_dfs.reactions.index.tolist()[1::])
+
+    return dogmatic_sbml_dfs
+
+
+def _connect_dogmatic_mappings(
+    species: str,
+    preferred_method: str = GENODEXITO_DEFS.BIOCONDUCTOR,
+    allow_fallback: bool = True,
+    r_paths: str | None = None,
+) -> dict:
+    """
+    Connect Dogmatic Mappings
+
+    Merge all ontologies into greedy clusters based on shared associations to entrez ids
+
+    Args:
+        species (str):
+            An organismal species (e.g., Homo sapiens)
+        r_paths (str or None)
+            Optional, p]ath to an R packages directory
+
+    Returns:
+        dict with:
+        - s_name_series: a series where the index is distinct molecular species and the values are names.
+        - cluster_consensus_identifiers_df: a pd.DataFrame where the index is distinct molecular species
+          and values are identifiers objects.
+    """
+
+    genodexito = Genodexito(
+        species=species,
+        preferred_method=preferred_method,
+        allow_fallback=allow_fallback,
+        r_paths=r_paths,
+    )
+
+    genodexito.create_mapping_tables(mappings=INTERCONVERTIBLE_GENIC_ONTOLOGIES)
+
+    genodexito.stack_mappings(ontologies=set(PROTEIN_ONTOLOGIES))
+    protein_mappings = genodexito.stacked_mappings
+
+    # apply greedy graph-based clustering to connect proteins with a common mapping to entrez
+    edgelist_df = utils.format_identifiers_as_edgelist(
+        protein_mappings, [IDENTIFIERS.ONTOLOGY, IDENTIFIERS.IDENTIFIER]
+    )
+    connected_indices = utils.find_weakly_connected_subgraphs(
+        edgelist_df[["ind", "id"]]
+    )
+
+    # add clusters to proteins. Each cluster will be a distinct molecular species
+    protein_mappings_w_clusters = protein_mappings.reset_index().merge(
+        connected_indices
+    )
+
+    # combine entrez + cluster so we can pass cluster to non-protein attributes
+    entrez_clusters = protein_mappings_w_clusters[
+        [ONTOLOGIES.NCBI_ENTREZ_GENE, "cluster"]
+    ].drop_duplicates()
+    # check for the other ontologies aside from proteins and entrez (since that's in the index)
+    other_ontologies = INTERCONVERTIBLE_GENIC_ONTOLOGIES.difference(
+        set(PROTEIN_ONTOLOGIES)
+    )
+    other_ontologies.discard(ONTOLOGIES.NCBI_ENTREZ_GENE)
+
+    genodexito.stack_mappings(ontologies=other_ontologies)
+    other_mappings = genodexito.stacked_mappings
+
+    other_mappings_w_clusters = entrez_clusters.merge(
+        other_mappings, left_on=ONTOLOGIES.NCBI_ENTREZ_GENE, right_index=True
+    )
+
+    possible_names = pd.concat(
+        [
+            protein_mappings_w_clusters.query("ontology in @NAME_ONTOLOGIES.keys()"),
+            other_mappings_w_clusters.query("ontology in @NAME_ONTOLOGIES.keys()"),
+        ]
+    )[["cluster", IDENTIFIERS.ONTOLOGY, IDENTIFIERS.IDENTIFIER]]
+
+    possible_names.loc[:, "ontology_preference"] = possible_names[
+        IDENTIFIERS.ONTOLOGY
+    ].map(NAME_ONTOLOGIES)
+
+    # remove possible names which are present in multiple clusters.
+    # all clusters will need unique names to use sbml_dfs_from_edgelist()
+    id_counts = (
+        possible_names[["cluster", IDENTIFIERS.IDENTIFIER]]
+        .drop_duplicates()
+        .value_counts(IDENTIFIERS.IDENTIFIER)
+    )
+    possible_names = possible_names[
+        ~possible_names[IDENTIFIERS.IDENTIFIER].isin(
+            id_counts[id_counts > 1].index.tolist()
+        )
+    ]
+
+    s_name_series = (
+        utils._add_nameness_score(possible_names, IDENTIFIERS.IDENTIFIER)
+        .sort_values(["ontology_preference", "nameness_score"])
+        .groupby("cluster")
+        .first()
+        .rename(columns={IDENTIFIERS.IDENTIFIER: SBML_DFS.S_NAME})[SBML_DFS.S_NAME]
+    )
+
+    protein_ids = protein_mappings_w_clusters.assign(bqb=BQB.IS)[
+        ["cluster", IDENTIFIERS.IDENTIFIER, IDENTIFIERS.ONTOLOGY, IDENTIFIERS.BQB]
+    ]
+    gene_ids = other_mappings_w_clusters.query("ontology in @GENE_ONTOLOGIES").assign(
+        bqb=BQB.IS_ENCODED_BY
+    )[["cluster", IDENTIFIERS.IDENTIFIER, IDENTIFIERS.ONTOLOGY, IDENTIFIERS.BQB]]
+    entrez_ids = entrez_clusters.assign(
+        ontology=ONTOLOGIES.NCBI_ENTREZ_GENE, bqb=BQB.IS_ENCODED_BY
+    ).rename(columns={ONTOLOGIES.NCBI_ENTREZ_GENE: IDENTIFIERS.IDENTIFIER})[
+        ["cluster", IDENTIFIERS.IDENTIFIER, IDENTIFIERS.ONTOLOGY, IDENTIFIERS.BQB]
+    ]
+
+    # combine all ids to setup a single cluster-level Identifiers
+    all_ids = pd.concat([protein_ids, gene_ids, entrez_ids])
+    all_ids.loc[:, IDENTIFIERS.URL] = [
+        identifiers.create_uri_url(x, y)
+        for x, y in zip(all_ids[IDENTIFIERS.ONTOLOGY], all_ids[IDENTIFIERS.IDENTIFIER])
+    ]
+
+    # create one Identifiers object for each new species
+    cluster_consensus_identifiers = {
+        k: identifiers.Identifiers(
+            list(
+                v[
+                    [
+                        IDENTIFIERS.ONTOLOGY,
+                        IDENTIFIERS.IDENTIFIER,
+                        IDENTIFIERS.URL,
+                        IDENTIFIERS.BQB,
+                    ]
+                ]
+                .reset_index(drop=True)
+                .T.to_dict()
+                .values()
+            )
+        )
+        for k, v in all_ids.groupby("cluster")
+    }
+
+    cluster_consensus_identifiers_df = pd.DataFrame(
+        cluster_consensus_identifiers, index=[SBML_DFS.S_IDENTIFIERS]
+    ).T
+    cluster_consensus_identifiers_df.index.name = "cluster"
+
+    out_dict = {
+        "s_name_series": s_name_series,
+        "cluster_consensus_identifiers_df": cluster_consensus_identifiers_df,
+    }
+
+    return out_dict