Virus Finding Tools: current solutions and limitations - Synthetic datasets

<p>The&nbsp;simulated RNA-seq datasets were&nbsp;built using FluxSimulator to generate paired-end reads. The first&nbsp;one contains thirty viral genomes from NCBI and the human genome (version GRCh38). The second&nbsp;dataset comprises five Human Rhinovirus A1, five Human&nbsp;Papillomavirus 16, and the human genome (version GRCh38).</p> <p>Concerning the first dataset, we chose the genomes to mix&nbsp;human, animal, and vegetable viruses for a fair analysis of each tool&#39;s reliability. Instead, the second dataset was created to&nbsp;understand the capability of each tool to assess the taxonomic&nbsp;level and the specificity of viral identification.</p> <p>We downloaded the genomes of the selected viruses from&nbsp;NCBI nucleotide, getting the FASTA file and the GFF3&nbsp;annotation. Next, we used AGAT&nbsp;to transform the GFF3&nbsp;files into GTF files. After that, we used FluxSimulator to&nbsp;generate a synthetic RNA-seq dataset for each species. Finally,&nbsp;we joined each simulated dataset to build a single RNA-seq&nbsp;sample. The number of reads generated for each species was&nbsp;chosen to act as a real sample.</p> <p>For each dataset, we include two compressed archives. The &quot;*_fastq_files.tar&quot; files contain&nbsp;the synthetic fastq files generated by FluxSimulator. The &quot;*_raw_results.tar&quot; files contain all the raw output produced by the tools employed in our benchmarking.</p> <p>The selected viruses for the first dataset are&nbsp;Human rhinovirus 1 strain ATCC VR-1559,&nbsp;Human Rhinovirus 3, Tomato mosaic virus,&nbsp;Molluscum contagiosum virus subtype 1,&nbsp;Apple mosaic virus RNA 3,&nbsp;Encephalomyocarditis virus,&nbsp;Human papillomavirus 52 isolate 52HB20,&nbsp;Hepatitis C virus genotype 1,&nbsp;Human papillomavirus type 31,&nbsp;Human papillomavirus type 54,&nbsp;JC&nbsp; polyomavirus,&nbsp;Marine RNA virus SF-2,&nbsp;Marine RNA virus JP-B,&nbsp;Hepatitis A virus,&nbsp;Human immunodeficiency virus 1,&nbsp;Anguillid herpes virus strain UK N080,&nbsp;Apis mellifera virus 14 isolate BFH508NG,&nbsp;Human enterovirus,&nbsp;Escherichia phage T7 isolate T7,&nbsp;Human herpesvirus 6B,&nbsp;Human measles virus,&nbsp;Cyprinid herpesvirus 3,&nbsp;Rotavirus C segment 8,&nbsp;Japanese encephalitis virus,&nbsp;Human papillomavirus 116,&nbsp;Influenza A virus&nbsp;(A/New York/392/2004(H3N2)) segment 4,&nbsp;Rotavirus RCU,&nbsp;Human parvovirus B19,&nbsp;Rous Sarcoma,&nbsp;Human papillomavirus 16.</p> <p>The selected strains and isolates for the second dataset are&nbsp;Human rhinovirus 1 strain ATCC VR-1559,&nbsp;Rhinovirus A1 strain 7A2,&nbsp;Rhinovirus A1 strain 5Q1,&nbsp;Rhinovirus A1 strain RvA1B/USA/2021/RJ9JKH,&nbsp;Rhinovirus A1 strain RvA1/USA/2021/L8MQLH,&nbsp;Human papillomavirus type 16,&nbsp;Human papillomavirus type 16 isolate 16CN37,&nbsp;Human papillomavirus type 16 isolate 16CN34,&nbsp;Human papillomavirus type 16 strain MML8,&nbsp;Human papillomavirus type 16 strain MML20.</p> <p>The tools employed in our benchmarking are&nbsp;VirusFinder, VirusSeq, VirTect,&nbsp;viGEN,&nbsp;VirDetect,&nbsp;DAMIAN,&nbsp;Metamap,&nbsp;Kraken2,&nbsp;Centrifuge.</p>