Skip to main content

Table 2 Summary metrics derived from using pure samples from Round 3 for baseline predictions

From: Summarizing performance for genome scale measurement of miRNA: reference samples and metrics

Measurement   Data Source miRNA Detected Log2 Signals AUCc Deviation
Process Technology by Round   Non- TrueNeg TruePos Lower Upper Range All Within Log2 Ratio Proportion
   Pure Mixes Reps Selective 1-to-1a Brain Placenta Limitb Limit Range Median IQR Sum
A NGS 3 3 3 928 56 103 104 5.8 20.0 14.3 0.959 0.996 -0.013 0.303 0.375
4 3 5.4 20.2 14.8 0.966 0.988 -0.006 0.311 0.358
5 3 6.3 20.0 13.8 0.899 0.938 -0.059 0.380 0.361
B NGS 3 3 3 868 56 134 47 7.3 20.2 12.8 0.905 0.953 -0.085 0.387 0.508
4 3 6.3 19.5 13.2 0.866 0.932 -0.092 0.399 0.448
5 3 6.7 19.7 12.9 0.848 0.904 -0.435 0.397 0.565
C NGS 3 3 3 421 35 29 49 5.3 16.3 11.0 0.944 0.971 0.030 0.266 0.332
4 3 11.1 16.1 5.0 0.831 0.850 0.040 0.626 0.460
5 3 15.7 15.7 0.0 0.848 NDd -0.124 0.839 0.556
D NGS 3 3 3 248 16 18 9 8.6 10.6 2.0 0.781 NDd -0.124 0.627 0.803
4 3 6.1 10.7 4.6 0.956 1.000 -0.178 0.539 0.645
E Digital Hybridization 3 3 3 693 65 8 24 7.1 13.3 6.1 0.947 1.000 -0.075 0.392 0.460
4 3 7.7 13.7 6.0 0.949 1.000 -0.260 0.495 0.532
5 3 13.2 14.4 1.1 0.925 NDd -0.257 0.680 0.579
  1. aLiver-selective miRNA or miRNA with approximately equal signals in both brain and placenta
  2. bLower limits for NGS and digital hybridization were calculated with at 95% of values falling within ± 0.585 log2 difference from mean
  3. cROC curves were prepared with P-values using three replicates of Mix1 and Mix2
  4. dNot determined (ND) due to insufficient number of true negative or true positive values