Skip to content

Performance

Benchmark runs used ViroSync commit a43a34f, the v1.0.7 core resources, one worker, and 16 threads. The benchmark disabled frameshift screening, TMVec2, InterProScan, and the optional structure steps. The figures come from benchmark commit 094b76e.

Select a figure to open the full-size image.

Synthetic boundary recovery

The synthetic set contains 60 loci and 3,365.2 kb of inserted sequence. Mean best-call boundary recall was 0.796 for ViroSync, 0.596 for ViralRecall v3.1.0, 0.307 for ViralRecall v2, 0.056 for DetectEVE v1.4.0, and 0.066 for EEfinder v1.1.1.

Synthetic boundary recovery and call burden

Boundary recovery and call burden across 60 synthetic loci. Panel a credits the call with the highest Jaccard index at each locus. Panel b compares missed and outside-truth sequence. Panel c partitions calls at a 500 bp one-to-one threshold. The test excludes 30 survey-derived loci whose source elements ViroSync found before this comparison. Region detection has no true negative, so the plot does not report specificity or accuracy.

Runtime and peak memory

Mean wall time across 30 SynEVEs-2 inputs was 143.7 seconds for ViroSync and 12.3 seconds for ViralRecall v3.1.0. Across five real-genome inputs, the means were 649.0 and 1,705.5 seconds. Campaign concurrency differed, so these values do not give a general speed ranking.

Runtime and peak resident memory

Per-input wall time and peak resident memory for 30 SynEVEs-2 inputs and five real-genome inputs. All tools used 16 threads or cores on the same 64-core host, but campaign concurrency differed. Wall time includes contention and database caching. Peak RSS is the largest single process, not the total for the process tree. EEfinder is absent because its campaign used a different load and concurrency.

Real-genome output

The five real genomes have no complete region-level truth set. The figure describes output burden and gene content, not detection accuracy.

Real-genome candidate burden and gene composition

Candidate count, called sequence, and gene composition across five real genomes. DetectEVE and EEfinder emit short homology hits. Their counts measure output granularity and are not direct locus-count comparisons.

See Methods for the ViroSync workflow and output rules.