Extended Data Fig. 1: MetaGraph API and search approaches.
From: Efficient and accurate search in petabase-scale sequence repositories
a) MetaGraph is designed to support a client-server infrastructure as exemplified here with a script in Python. In a few steps, several remote (or local) graph instances can be created and queried interactively. Results are returned as a data frame that can be used for further analyses. b) Conceptual overview of remote or local index distribution. Every graph index runs on a separate server, accepting queries via the client API. c-e) Schematic representation of two main approaches for sequence search. c) Counting exact k-mer matches between query and graph. d) Alignment finds all closest paths within a given edit distance. e) Batched sequence search retrieves a decompressed subgraph (query graph) from the full compressed annotated graph for subsequent query. All query sequences are combined into an intermediate batch graph, which is then traversed to extract contigs to be queried against the full index. Hits and their corresponding annotations are aggregated to construct the final query graph, which is then searched against with the original query sequences.
