Fernando Lopez, Quanam | Dataworks Summit EU 2018
summary
Joint project with GeneLifes builds a production bioinformatics pipeline that scores genomic variants for pathogenicity using heuristics, ingesting about 3 million abstracts (roughly 300/day) for literature-driven relationships. Architecture uses Hortonworks HDP (HDFS, Spark) for scalable storage compute, Neo4j for a semantic graph with millions of nodes and edges, NLP from Nine, and Solr for search. Data volumes approach 1 TB per patient. The system is deployed and in active use with privacy and GDPR concerns addressed.
AI-written summary · about
