napistu 0.1.0__py3-none-any.whl

napistu/ingestion/yeast.py

@@ -0,0 +1,147 @@
+from __future__ import annotations
+
+import pandas as pd
+from napistu import identifiers
+from napistu import sbml_dfs_core
+from napistu import source
+from napistu import utils
+from napistu.constants import BQB
+from napistu.constants import MINI_SBO_FROM_NAME
+from napistu.ingestion.constants import YEAST_IDEA_KINETICS_URL
+from napistu.ingestion.constants import YEAST_IDEA_PUBMED_ID
+from napistu.ingestion.constants import YEAST_IDEA_SOURCE
+from napistu.ingestion.constants import YEAST_IDEA_TARGET
+
+
+def download_idea(output_dir: str) -> None:
+    # save to
+    utils.download_and_extract(YEAST_IDEA_KINETICS_URL, output_dir)
+
+    # TODO: since only a single file is outputted, it makes sense to download and extract the data, then copy it to a target URI
+    # TODO: add GCS support
+    pass
+
+
+def convert_idea_kinetics_to_sbml_dfs(
+    idea_path: str,
+) -> sbml_dfs_core.SBML_dfs:
+    """
+    Convert IDEA Kinetics to SBML DFs
+
+    Format yeast induction regulator->target relationships as a directed graph.
+
+    Args:
+        idea_path: Path to the IDEA Kinetics file.
+
+    Returns:
+        SBML_dfs: an SBML_dfs object containing molecular species and their interactions.
+        Kinetic attributes are included as reactions_data.
+
+    """
+
+    # TO DO - replace with GCS support (currently this just reads a local .tsv)
+    idea_kinetics_df = pd.read_csv(idea_path, sep="\t")
+
+    # separate based on whether the change is probably direct or indirect
+    idea_kinetics_df["directness"] = [
+        "direct" if t_rise < 15 else "indirect" for t_rise in idea_kinetics_df["t_rise"]
+    ]
+
+    # reduce cases of multiple TF-target pairs to a single entry
+    distinct_edges = (
+        idea_kinetics_df.groupby([YEAST_IDEA_SOURCE, YEAST_IDEA_TARGET], as_index=True)
+        .apply(_summarize_idea_pairs)
+        .reset_index()
+    )
+
+    # add some more fields are reformat
+    formatted_distinct_edges = distinct_edges.rename(
+        {YEAST_IDEA_SOURCE: "upstream_name", YEAST_IDEA_TARGET: "downstream_name"},
+        axis=1,
+    ).assign(
+        upstream_compartment="cellular_component",
+        downstream_compartment="cellular_component",
+        # tag reactions with the IDEA publication
+        r_Identifiers=identifiers._format_Identifiers_pubmed(YEAST_IDEA_PUBMED_ID),
+        r_isreversible=False,
+    )
+
+    # create some nice interaction names before we rename the roles as SBO terms
+    formatted_distinct_edges["r_name"] = [
+        f"{u} {d} {r} of {t}"
+        for u, d, r, t in zip(
+            formatted_distinct_edges["upstream_name"],
+            formatted_distinct_edges["directness"],
+            formatted_distinct_edges["role"],
+            formatted_distinct_edges["downstream_name"],
+        )
+    ]
+
+    # final interaction output
+    # replace readable roles with entries in the SBO ontology
+    interaction_edgelist = formatted_distinct_edges.replace(
+        {"role": MINI_SBO_FROM_NAME}
+    ).rename({"role": "sbo_term"}, axis=1)
+
+    species_df = pd.DataFrame(
+        {
+            "s_name": list(
+                {
+                    *idea_kinetics_df[YEAST_IDEA_SOURCE],
+                    *idea_kinetics_df[YEAST_IDEA_TARGET],
+                }
+            )
+        }
+    )
+
+    # create Identifiers objects for each species
+    species_df["s_Identifiers"] = [
+        identifiers.Identifiers(
+            [{"ontology": "gene_name", "identifier": x, "bqb": BQB.IS}]
+        )
+        for x in species_df["s_name"]
+    ]
+
+    # Constant fields (for this data source)
+
+    # setup compartments (just treat this as uncompartmentalized for now)
+    compartments_df = sbml_dfs_core._stub_compartments()
+
+    # Per convention unaggregated models receive an empty source
+    interaction_source = source.Source(init=True)
+
+    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,
+        # additional attributes (directness) are added to reactions_data
+        keep_reactions_data="idea",
+    )
+    sbml_dfs.validate()
+
+    return sbml_dfs
+
+
+def _summarize_idea_pairs(pairs_data: pd.DataFrame) -> pd.Series:
+    """Rollup multiple records of a TF->target pair into a single summary."""
+
+    # specify how to aggregate results if there are more than one entry for a TF-target pair
+    # pull most attributes from the earliest change
+    # this will favor direct over indirect naturally
+    earliest_change = pairs_data.sort_values("t_rise").iloc[0].to_dict()
+
+    KEYS_SUMMARIZED = ["v_inter", "v_final", "t_rise", "t_fall", "rate", "directness"]
+    kinetic_timing_dict = {k: earliest_change[k] for k in KEYS_SUMMARIZED}
+
+    # map v_inter (log2 fold-change change following perturbation) onto SBO terms for interactions
+    if (any(pairs_data["v_inter"] > 0)) and (any(pairs_data["v_inter"] < 0)):
+        kinetic_timing_dict["role"] = "modifier"
+    elif all(pairs_data["v_inter"] > 0):
+        kinetic_timing_dict["role"] = "stimulator"
+    elif all(pairs_data["v_inter"] < 0):
+        kinetic_timing_dict["role"] = "inhibitor"
+    else:
+        ValueError("Unexpected v_inter values")
+
+    return pd.Series(kinetic_timing_dict)