Virus Finding Tools: current solutions and limitations - Real Datasets

We assessed the behavior of 9 virus-finding tools using real data. We used six datasets from which we took 3 to 5 samples. In each dataset, at least one virus was present. Confirmation of the virus was performed by RNA-seq for some samples. In the other datasets, the virus was experimentally added to the cellular sample. Only the confirmed species was used to test the reliability of the methods, whereas the others were discarded since no experimental validation was available. Each dataset was picked from GEO and downloaded through the ENA Browser.

We employed the following datasets: GEO GSE74949 (Samples: SRR2932830, SRR2932844, SRR2932845, SRR2932846, SRR2932847 - Human Papillomavirus 16), GEO GSE65486 (Samples: SRR1946683, SRR1946684, SRR1946685, SRR1946686, SRR1946687 - Hepatitis B Virus), GEO GSE61141 (Samples: SRR1565938, SRR1565939, SRR1565940, SRR1565941, SRR1565942 - Human Rhinovirus 16), GEO GSE59717 (Samples: SRR1523654, SRR1523655, SRR1523656, SRR1523657, SRR1523668 - Human alphaherpesvirus 1), GEO GSE148729 (Samples: SRR11550056, SRR11550033 - Severe Acute Respiratory Syndrome Coronavirus 2), GEO GSE84346 (Samples: SRR3898707, SRR3898708, SRR3898709 - Hepatitis C virus), ENA PRJNA257197 (Samples: SRR1553464 - Zaire Ebolavirus)

The tools employed in our benchmarking are VirusFinder, VirusSeq, VirTect, viGEN, VirDetect, DAMIAN, Metamap, Kraken2, Centrifuge. DAMIAN could not be used for single-ended samples (GEO GSE84346).

For each sample, we provide a ".tar.bz2" archive. Inside each archive, there is one folder for each analysis, as described in our manuscript. Due to large file size, FASTQ and intermediate analysis files are not included.