Implement DepMap source

README.md

@@ -40,6 +40,7 @@ non-causal contextual relations including properties, ontology, and data.
 | [PubMed](https://ncbi.nlm.nih.gov/)                                 | published_in                        | The publication represented by the source is published in the journal represented by the target.                                                   |
 | [CellMarker](http://xteam.xbio.top/)                                | has_marker                          | The cell type represented by the source node has the gene marker represented by the target node.                                                   |
 | [InterPro](https://www.ebi.ac.uk/interpro/)                         | has_domain                          | The gene represented by the source node has a protein domain represented by the target node.                                                       |
+| [DepMap](https://depmap.org/portal/) | codependent_with | The gene represented by the source node is codependent with the gene represented by the target node in cancer cell lines.                          |
 
 ## Installation
 

src/indra_cogex/sources/depmap/__init__.py

@@ -0,0 +1,82 @@
+"""Process DepMap, a resource for gene-gene dependencies in cancer cell lines."""
+
+import logging
+import pickle
+import tqdm
+from collections import defaultdict
+
+import click
+import pystow
+from indra.databases import hgnc_client
+
+from indra_cogex.representation import Node, Relation
+from indra_cogex.sources.processor import Processor
+
+
+__all__ = [
+    "DepmapProcessor",
+]
+
+logger = logging.getLogger(__name__)
+
+SUBMODULE = pystow.module("indra", "cogex", "depmap")
+
+DEPMAP_SIGS = pystow.join('depmap_analysis', 'depmap', '21q2',
+                          name='dep_stouffer_signif.pkl')
+
+
+def load_sigs():
+    with open(DEPMAP_SIGS, 'rb') as f:
+        df = pickle.load(f)
+    sig_by_gene = defaultdict(dict)
+    for row in tqdm.tqdm(df.itertuples(), total=len(df),
+                         desc='Processing DepMap significant pairs'):
+        a, b = sorted(row.Index)
+        a_hgnc_id = hgnc_client.get_current_hgnc_id(a)
+        b_hgnc_id = hgnc_client.get_current_hgnc_id(b)
+        if a_hgnc_id is None or b_hgnc_id is None:
+            continue
+        if isinstance(a_hgnc_id, list):
+            a_hgnc_id = a_hgnc_id[0]
+        if isinstance(b_hgnc_id, list):
+            b_hgnc_id = b_hgnc_id[0]
+        a_current = hgnc_client.get_hgnc_name(a_hgnc_id)
+        b_current = hgnc_client.get_hgnc_name(b_hgnc_id)
+        sig_by_gene[(a_current, a_hgnc_id)][(b_current, b_hgnc_id)] = row.logp
+    sigs_by_gene = dict(sig_by_gene)
+    return sigs_by_gene
+
+
+class DepmapProcessor(Processor):
+    """Processor for the DepMap dataset."""
+
+    name = "depmap"
+    node_types = ["BioEntity"]
+    depmap_relation = "codependent_with"
+
+    def __init__(self):
+        """Initialize the DisGeNet processor."""
+        self.sigs_by_gene = load_sigs()
+
+    def get_nodes(self):  # noqa:D102
+        all_genes = set(self.sigs_by_gene)
+        for genes in self.sigs_by_gene.values():
+            all_genes |= set(genes)
+
+        for gene_name, hgnc_id in all_genes:
+            yield Node(db_ns="HGNC", db_id=hgnc_id, labels=["BioEntity"],
+                       data={'name': gene_name})
+
+    def get_relations(self):  # noqa:D102
+        for (a, a_hgnc_id), genes in \
+                tqdm.tqdm(self.sigs_by_gene.items(),
+                          desc='Processing DepMap into relations'):
+            for (b, b_hgnc_id), logp in genes.items():
+                yield Relation(
+                    source_ns="HGNC",
+                    source_id=a_hgnc_id,
+                    target_ns="HGNC",
+                    target_id=b_hgnc_id,
+                    rel_type=self.depmap_relation,
+                    data={"logp": logp},
+                )

src/indra_cogex/sources/depmap/__main__.py

@@ -0,0 +1,5 @@
+from . import DepmapProcessor
+
+
+if __name__ == "__main__":
+    DepmapProcessor.cli()