Gauge your phage: benchmarking of bacteriophage identification tools in metagenomic sequencing data

Microbiome

Table 2 Number of sequences at each stage of the RefSeq benchmarking workflow

Dataset	Chromosome (n)	Plasmid (n)	Phage (n)	Chromosome (bp)	Plasmid (bp)	Phage (bp)
1. Post-2020	7400	9960	1849	30,034,515,475	1,009,997,498	128,660,045
2. Post-2020 dereplicated	3546	4453	901	14,845,391,115	480,796,703	62,387,054
3. Subsampled	253	309	901	1,011,740,231	28,782,822	62,387,054
4. Prophage removed	2307 (2088)	400 (91)	901	965,959,872	27,323,488	62,387,054
5. pVOG removed	2065	313	901	848,361,160	24,433,971	62,387,054
6. Fragmented artificial contigs	104,003	2754	6664	830,889,456	21,783,942	53,426,665

Columns labelled with (n) contain the number of sequences at each step, and columns with (bp) indicate the number of base pairs at each step. Steps are numbered as follows:
1. Sequences downloaded from RefSeq which were deposited between 1 January 2020 and 12 August 2021
2. Sequences from (1) which were then dereplicated with RefSeq sequences deposited before 1.st January 2020 and training sets for DeepVirFinder, Seeker, VIBRANT, VirFinder, and VirSorter2
3. Host sequences (chromosome and plasmids) from (2) which were subsampled by a factor of 14.3
4. Host sequences from (3) with prophage removal using Phigaro and PhageBoost. The number in parentheses indicates the number of prophages removed.
5. Host sequences from (4) with sequences that have ≥ 30% of their open reading frames having hits to the pVOG database removed
6. All sequences from (5) randomly and uniformly fragmented to sizes between 1 and 15 kbp for use in the benchmarking study

ISSN: 2049-2618