Federated SPARQL query performance evaluation for exploring disease model mouse: combining gene expression, orthology, and disease knowledge graphs

Table 7 Comparison of the runtimes of the federated approach from the BioResource MetaDB (Tsukuba) to the Bgee (Lausanne), and the centralized approach at Tsukuba and Lausanne

Query approach	Federated or Centralized	No. of retrieved genes	Mean of runtime	[A–B]	[C–B]
Federated (Tsukuba to Lausanne): Example 7	Federated	42,448	48 s [A]	37 s
Centralized (Lausanne to Lausanne)*	Centralized	42,448	11 s [B]	37 s	3 s
Centralized (Tsukuba to Tsukuba): Example 9	Centralized	42,448	14 s [C]		3 s

We conducted the performance by executing the SPARQL query for Bgee data (Examples 7 and 9). The executed query includes the query conditions: the prefrontal cortex (UBERON:0000451) as location of gene expression, a high confidence level for expression data, the sex condition for “any sex type”
*: After we removed a row including the SERVICE keyword from Example 7 and executed it
A–B: Data transfer time between Tsukuba (RIKEN BRC) and Lausanne (Bgee)
C–B: The difference in the retrieval time of the BioResource MetaDB and Bgee

ISSN: 1472-6947