Skip to main content

Table 8 Scores obtained for the primary dataset using cross validation

From: Environmental metagenome classification for constructing a microbiome fingerprint

Ryan [21]#correct710252060161820\(\sum =176\)
Sanchez et al. [24]#correct9111101760341720\({\sum }=278\)
Harris et al. [32]Ns=N/A
Walker and Datta [22]TPR (median)0.60.620.580.950.870.760.30.7Ns=211
Zhu [25]#correct531141451311715\({\sum }=250\)
Chierici et al. [23]Ns=311
Our method using Mash#correct1515502060311920\({\sum }=230\)
sketch size=1000PPV0.340.261.000.671.
Our method using Mash#correct1516422060342020\({\sum }=227\)
sketch size=10000PPV0.650.181.000.831.
Our method using Mash#correct1516442060341920\({\sum }=228\)
sketch size=100000PPV0.600.
Our method using CoMeta#correct4121162037341320\({\sum }=256\)
(class-level filtering)PPV0.670.630.920.741.000.9710.42Ns=311
Our method using CoMeta#correct4131132057301619\({\sum }=272\)
(sample-level filtering)PPV0.570.650.920.650.921.000.940.9Ns=311
  1. We report the number of correctly classified samples (#correct), precision (PPV), and recall (TPR) for each class, as well as the overall accuracy (ACC). Some of the values are missing, as they were not reported in the referenced papers. Also, we show the number of samples (Ns), as in some works, the results for a subset of all of Ns=311 samples were reported