Table 1.

Benchmarking test description

Test No.	Aims	Methods
1	Assess evolution of tool speed performance. Involved Gigwa v1, Gigwa v2, VCFtools v0.1.13 (originally benchmarked) [4], and VCFtools v0.1.16 (latest at assessment time)	Run on configuration 1 using dataset 1 (along with sub-sampled versions, so as to obtain 6 different databases), all with the same number of individuals (i.e., 3,000) but with various numbers of markers. Query was a MAF range between 10% and 30% applied to the first 2,000 individuals
2	(i) Assess performance of latest versions of tools (Gigwa v2 and VCFtools v0.1.16) when simultaneously querying on variant-level (indexed in Gigwa) and genotype-level (unindexed in Gigwa) fields. (ii) Estimate the benefit of migrating to high-performance hardware by monitoring differences in response times between tools	Run on configuration 2 using dataset 1 without its derivatives, sub-sampling now being performed on the fly by restricting the search to a varying list of chromosomes. The query was the same MAF range query as above
3	(i) Test Gigwa v2’s suitability for working on very large datasets. (ii) Compare trends with those observed in a small dataset (Test 2)	Run on configuration 2 using dataset 2, sub-sampling being performed on the fly by restricting the search to a varying list of chromosomes. The query was the same MAF range query as above