SemKraph KG: Pipeline Output Knowledge Graphs

SemKraph KG: Pipeline Output Knowledge Graphs Knowledge graphs produced by the semkraphkg-pipeline from the SemKraph KG: Pipeline Input Datasets deposit: the SemKraph KG (normalized SemMedDB + iKraph, general medicine) and the SemKraph CM Subgraph (Complementary Medicine seeds + 1-hop neighborhood). Node identifiers are normalized through the Translator NodeNorm service and edge predicates are mapped to the Biolink Model. About the Project COMBINI (Connecting Complementary Medicine and Biological Knowledge to support Integrative Health) is a multi-institutional project (University of Illinois/NCSA, University of Minnesota, Mayo Clinic) funded by the NIH/NCCIH. The objective of the project is to develop informatics methods and resources that consolidate evidence on complementary medicine, integrate it with conventional medicine in an AI-ready format, and enable knowledge management and scientific discovery in complementary and integrative health. Inputs: SemMedDB (NLM semantic predications from PubMed), iKraph (biomedical entity/relation graph), and the COMBO ontology (SKOS-enriched complementary medicine concepts). See the input deposit for the exact source files. Files Included Zenodo stores a flat file list, so directory structure from the pipeline output tree is encoded in the filename prefix. Large tabular files are uploaded gzip-compressed. Component File prefix Description SemKraph KG (Neo4j-import-ready) gen_med_graph_neo4j_* Normalized general-medicine KG as Neo4j bulk-import CSVs: semmed_nodes_normalized.csv (~330K nodes), semmed_edges_normalized.csv (~19.4M edges), ikraph_nodes_normalized.csv (~308K nodes), ikraph_edges_normalized.csv (~6.1M edges) SemKraph KG (KGX) kgx_gen_med_* KGX-format TSV export of the general-medicine KG: nodes.tsv (~598K nodes), edges.tsv (~25.4M edges) SemKraph CM Subgraph (KGX) kgx_cm_subgraph_* KGX-format TSV export of the CM subgraph: nodes.tsv (~26K nodes), edges.tsv (~111K edges) SemKraph CM Subgraph (detail) cm_subgraph_* Normalized CM subgraph tables: COMBO seed nodes, seed+1-hop and seed-induced SemMedDB/iKraph edge sets (*_combo_seeds_and_1hop, *_combo_seeds_only), the CM concept table, COMBO normalization links, plus combo_subgraph_summary.json with build statistics How to Use Decompress and load the KGX exports, or bulk-import the Neo4j CSVs: gunzip kgx_gen_med_nodes.tsv.gz kgx_gen_med_edges.tsv.gz neo4j-admin database import full \ --nodes=gen_med_graph_neo4j_semmed_nodes_normalized.csv \ --relationships=gen_med_graph_neo4j_semmed_edges_normalized.csv \ neo4j To regenerate these outputs from the input deposit, see the pipeline repository: pip install -r requirements.txt python3 cli.py download --semmed-dir data/semmed --ikraph-dir data/ikraph \ --combo-dir data/combo --reltype-xlsx ikraph_reltype_to_biolink_mapping_review.xlsx python3 cli.py transform --output-dir output/kg_build_out ... python3 cli.py merge --output-dir output/kg_build_out ...