Fig. 6

CLIMB-COVID inbound data architecture diagram. Local sites (grey, top left) generate consensus FASTA and alignment BAM for each sequenced sample. Corresponding metadata is collected and managed into a CSV using the consortium template. FASTA and BAM are uploaded to CLIMB using scp or rsync. Metadata is converted from CSV to JSON by the metadata uploader tool and passed to the Majora API to be processed (purple, right). The Elan inbound pipeline (green, left) queries the Majora metadata database using the Ocarina command line client (yellow, right). Elan matches Majora metadata to uploaded files on CLIMB (blue, left) and conducts quality control. Quality metrics are passed to Majora through Ocarina. Outbound distribution pipelines (orange, centre) are able to query Majora using Ocarina and package high-quality sequences for GISAID. ENA and INSDC databases (red). Downstream pipelines (green, centre) such as Datapipe and Phylopipe generate alignments, trees and other analysis artifacts that are shared within the consortium and made publicly available via CLIMB-COVID’s S3 storage (red). Programmes in yellow boxes indicate software built specifically for CLIMB-COVID