Skip to main content

Table 3 Performance comparison between ENVirT, PHACCS and CatchAll on simulated contig spectra

From: ENVirT: inference of ecological characteristics of viruses from metagenomic data

Input parameters (expected result) ENVirT PHACCS CatchAll
L 0 M 0 T 0 d 0 Evenness f max M T d S min M T d S min M
12500 300 exp 0.030 0.790 2.956% 300 exp 0.030 0.00x10 0 4096 exp 0.030 1.37x10 -3 2829.6 p
12500 1000 log 0.900 0.995 0.661% 1000 log 0.900 0.00x10 0 1000 log 0.900 0.00x10 0 92628.3 c
12500 5000 lgn 2.500 0.655 11.849% 5000 lgn 2.500 0.00x10 0 23563 pl 1.313 1.01x10 4 3246.1 p
12500 10000 pl 0.700 0.913 1.997% 10000 pl 0.700 0.00x10 0 10000 pl 0.700 0.00x10 0 696.3 p
50000 300 exp 0.030 0.790 2.956% 300 exp 0.030 0.00x10 0 10000 exp 0.030 4.31x10 -4 15712.6 p
50000 1000 log 0.900 0.995 0.661% 1000 log 0.900 0.00x10 0 1000 log 0.900 0.00x10 0 n/a
50000 5000 lgn 2.500 0.655 11.849% 5000 lgn 2.500 0.00x10 0 4996 lgn 2.500 1.78x10 -3 799.8 p
50000 10000 pl 0.700 0.913 1.997% 10000 pl 0.700 0.00x10 0 10000 pl 0.700 0.00x10 0 413688.9 c
125000 300 exp 0.030 0.790 2.956% 300 exp 0.030 0.00x10 0 10000 exp 0.060 1.87x10 -4 70340.9 c
125000 1000 log 0.900 0.995 0.661% 1000 log 0.900 0.00x10 0 1000 log 0.900 0.00x10 0 n/a
125000 5000 lgn 2.500 0.655 11.849% 5000 lgn 2.500 0.00x10 0 5000 lgn 2.500 0.00x10 0 2303.2 p
125000 10000 pl 0.700 0.913 1.997% 10000 pl 0.700 0.00x10 0 10000 pl 0.700 0.00x10 0 n/a
300000 300 exp 0.030 0.790 2.956% 300 exp 0.030 0.00x10 0 4096 exp 0.030 7.92x10 -5 160243.9 c
300000 1000 log 0.900 0.995 0.661% 1000 log 0.900 0.00x10 0 1000 log 0.900 0.00x10 0 n/a
300000 5000 lgn 2.500 0.655 11.849% 5000 lgn 2.500 0.00x10 0 5000 lgn 2.500 0.00x10 0 146552.7 c
300000 10000 pl 0.700 0.913 1.997% 8547 pl 0.689 3.00x10 -3 10000 pl 0.700 0.00x10 0 n/a
  1. Contig spectra were generated with parameters: R=10000, r= 100bp and o= 35bp. Both ENVirT and PHACCS were provided with the true average genome length (L0) value. pl = power-law distribution, exp = exponential distribution, log = logarithmic distribution and lgn = lognormal distribution. Smin = the value of the cost function corresponding to the estimated values of M,T and d for each method. For each spectrum, the CatchAll estimate having the minimum error compared to M0 is reported. p = best discounted parametric model produced by CatchAll. c = Chao1 non-parametric estimate. n/a denotes samples for which CatchAll failed to produce an output