CompaRNA - on-line benchmarks of RNA structure prediction methods
Home

Methods
Datasets
Rankings
RNA 2D Atlas

Help
FAQ

Contact us
RSS feeds
Twitter

Table of contents:

  1. Overview

  2. Performance Plots

  3. Performance of ProbKnot - scored higher in this pairwise comparison

  4. Performance of Mastr(20) - scored lower in this pairwise comparison

  5. Compile and download dataset for ProbKnot & Mastr(20) [.zip] - may take several seconds...


Overview

Metric ProbKnot Mastr(20)
MCC 0.597 > 0.483
Average MCC ± 95% Confidence Intervals 0.609 ± 0.110 > 0.450 ± 0.179
Sensitivity 0.536 > 0.297
Positive Predictive Value 0.669 < 0.791
Total TP 361 > 200
Total TN 154676 < 154963
Total FP 218 > 61
Total FP CONTRA 30 > 6
Total FP INCONS 149 > 47
Total FP COMP 39 > 8
Total FN 313 < 474
P-value 2.06323189071e-08

^top




Performance plots


  1. Comparison of performance of ProbKnot and Mastr(20). Positive Predictive Value (PPV) is plotted against sensitivity. Each dot represents a single test of each method. See tables below for raw data (individual counts for ProbKnot and Mastr(20)).

  2. Average Matthews Correlation Coefficients (MCC) with 95% confidence intervals (CIs) were plotted for different RNA families, for which at least 3 members were present in the benchmarking dataset. 'n' denotes the number of MCCs used to calculate the average and CI. See tables below for raw data (individual counts for ProbKnot and Mastr(20)).

  3. Comparison of average Matthews Correlation Coefficients (MCCs) for ProbKnot and Mastr(20). The whiskers correspond to 95% confidence intervals (CIs). 'n' denotes the number of MCCs used to calculate average MCCs and CIs. See tables below for raw data (individual counts for ProbKnot and Mastr(20)).

^top





Performance of ProbKnot - scored higher in this pairwise comparison

1. Total counts & total scores for ProbKnot

Total Base Pair Counts
Total TP 361
Total TN 154676
Total FP 218
Total FP CONTRA 30
Total FP INCONS 149
Total FP COMP 39
Total FN 313
Total Scores
MCC 0.597
Average MCC ± 95% Confidence Intervals 0.609 ± 0.110
Sensitivity 0.536
Positive Predictive Value 0.669
Nr of predictions 15

^top



2. Individual counts for ProbKnot [ download as .csv ]

RNA Chain Rfam family MCC SENS PPV TP TN FP FP CONTRA FP INCONS FP COMP FN
2XQD_Y 0.90 0.81 1.00 22 2828 1 0 0 1 5
3AMU_B 0.65 0.59 0.73 16 2981 8 0 6 2 11
3IZ4_A 0.52 0.46 0.60 61 70774 46 6 35 5 71
3IZF_C 0.72 0.61 0.85 33 6864 6 0 6 0 21
3NPB_A 0.72 0.61 0.85 28 6988 8 1 4 3 18
3O58_3 0.31 0.34 0.29 12 12362 41 4 25 12 23
3O58_2 0.76 0.76 0.76 29 7222 10 3 6 1 9
3PDR_A 0.74 0.64 0.85 46 12826 10 1 7 2 26
3RKF_A 0.73 0.59 0.91 20 2189 2 1 1 0 14
3SD1_A 0.55 0.48 0.65 20 3885 11 2 9 0 22
4A1C_2 0.13 0.15 0.12 5 11738 49 7 31 11 28
4A1C_3 0.73 0.61 0.87 33 7102 6 1 4 1 21
4AOB_A 0.50 0.40 0.63 17 4344 11 2 8 1 25
4ENB_A 0.73 0.58 0.92 11 1263 1 1 0 0 8
4ENC_A 0.45 0.42 0.50 8 1310 8 1 7 0 11

^top



Performance of Mastr(20) - scored lower in this pairwise comparison

1. Total counts & total scores for Mastr(20)

Total Base Pair Counts
Total TP 200
Total TN 154963
Total FP 61
Total FP CONTRA 6
Total FP INCONS 47
Total FP COMP 8
Total FN 474
Total Scores
MCC 0.483
Average MCC ± 95% Confidence Intervals 0.450 ± 0.179
Sensitivity 0.297
Positive Predictive Value 0.791
Nr of predictions 15

^top



2. Individual counts for Mastr(20) [ download as .csv ]

RNA Chain Rfam family MCC SENS PPV TP TN FP FP CONTRA FP INCONS FP COMP FN
2XQD_Y 0.88 0.78 1.00 21 2829 0 0 0 0 6
3AMU_B 0.77 0.59 1.00 16 2987 1 0 0 1 11
3IZ4_A 0.00 0.00 0.00 0 70876 0 0 0 0 132
3IZF_C 0.73 0.63 0.85 34 6863 7 1 5 1 20
3NPB_A 0.30 0.24 0.39 11 6993 18 1 16 1 35
3O58_3 0.00 0.00 0.00 0 12403 0 0 0 0 35
3O58_2 0.76 0.76 0.76 29 7222 12 3 6 3 9
3PDR_A 0.00 0.00 0.00 0 12880 0 0 0 0 72
3RKF_A 0.70 0.50 1.00 17 2194 0 0 0 0 17
3SD1_A 0.61 0.50 0.75 21 3888 7 1 6 0 21
4A1C_2 0.00 0.00 0.00 0 11781 0 0 0 0 33
4A1C_3 0.70 0.59 0.84 32 7102 7 0 6 1 22
4AOB_A 0.39 0.26 0.58 11 4352 9 0 8 1 31
4ENB_A 0.39 0.16 1.00 3 1272 0 0 0 0 16
4ENC_A 0.51 0.26 1.00 5 1321 0 0 0 0 14

^top


Matthews Correlation Coeffient, Sensitivity and Positive Predictive Value have been calculated based on the paper by Gardener & Giegerich, 2004.