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 Pknots - scored higher in this pairwise comparison

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

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


Overview

Metric Pknots RSpredict(20)
MCC 0.670 > 0.625
Average MCC ± 95% Confidence Intervals 0.705 ± 0.160 > 0.612 ± 0.179
Sensitivity 0.691 > 0.557
Positive Predictive Value 0.658 < 0.712
Total TP 248 > 200
Total TN 26350 < 26446
Total FP 165 > 95
Total FP CONTRA 43 > 38
Total FP INCONS 86 > 43
Total FP COMP 36 > 14
Total FN 111 < 159
P-value 1.74172190343e-08

^top




Performance plots


  1. Comparison of performance of Pknots and RSpredict(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 Pknots and RSpredict(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 Pknots and RSpredict(20)).

  3. Comparison of average Matthews Correlation Coefficients (MCCs) for Pknots and RSpredict(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 Pknots and RSpredict(20)).

^top





Performance of Pknots - scored higher in this pairwise comparison

1. Total counts & total scores for Pknots

Total Base Pair Counts
Total TP 248
Total TN 26350
Total FP 165
Total FP CONTRA 43
Total FP INCONS 86
Total FP COMP 36
Total FN 111
Total Scores
MCC 0.670
Average MCC ± 95% Confidence Intervals 0.705 ± 0.160
Sensitivity 0.691
Positive Predictive Value 0.658
Nr of predictions 14

^top



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

RNA Chain Rfam family MCC SENS PPV TP TN FP FP CONTRA FP INCONS FP COMP FN
3A2K_C 0.50 0.55 0.48 12 1083 13 3 10 0 10
3GX2_A 0.55 0.57 0.55 16 1420 14 4 9 1 12
3IVN_B 0.91 0.87 0.95 20 882 1 0 1 0 3
3JYV_7 0.97 0.95 1.00 19 1092 2 0 0 2 1
3LA5_A 0.94 0.88 1.00 22 932 0 0 0 0 3
3NPB_A 0.84 0.81 0.88 30 2244 9 0 4 5 7
3O58_3 0.32 0.45 0.24 10 4722 39 18 14 7 12
3PDR_A 0.64 0.64 0.65 32 4791 19 4 13 2 18
3RKF_A 0.91 0.88 0.95 21 844 1 0 1 0 3
3SD1_A 0.78 0.76 0.81 22 1506 5 1 4 0 7
4A1C_2 0.33 0.40 0.29 8 4488 36 9 11 16 12
4AOB_A 0.19 0.21 0.21 6 1409 23 3 19 1 23
4ENB_A 1.00 1.00 1.00 15 457 2 0 0 2 0
4ENC_A 0.97 1.00 0.94 15 480 1 1 0 0 0

^top



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

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

Total Base Pair Counts
Total TP 200
Total TN 26446
Total FP 95
Total FP CONTRA 38
Total FP INCONS 43
Total FP COMP 14
Total FN 159
Total Scores
MCC 0.625
Average MCC ± 95% Confidence Intervals 0.612 ± 0.179
Sensitivity 0.557
Positive Predictive Value 0.712
Nr of predictions 14

^top



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

RNA Chain Rfam family MCC SENS PPV TP TN FP FP CONTRA FP INCONS FP COMP FN
3A2K_C 0.88 0.86 0.90 19 1087 2 0 2 0 3
3GX2_A 0.40 0.21 0.75 6 1441 2 1 1 0 22
3IVN_B 0.91 0.87 0.95 20 882 1 1 0 0 3
3JYV_7 0.97 0.95 1.00 19 1092 2 0 0 2 1
3LA5_A 0.89 0.80 1.00 20 934 0 0 0 0 5
3NPB_A -0.01 0.00 0.00 0 2272 6 1 5 0 37
3O58_3 0.38 0.50 0.30 11 4727 29 15 11 3 11
3PDR_A 0.80 0.70 0.92 35 4802 5 1 2 2 15
3RKF_A 0.91 0.88 0.95 21 844 1 1 0 0 3
3SD1_A 0.82 0.86 0.78 25 1501 7 5 2 0 4
4A1C_2 0.25 0.30 0.21 6 4487 30 12 11 7 14
4AOB_A 0.33 0.21 0.55 6 1426 5 1 4 0 23
4ENB_A 0.54 0.40 0.75 6 464 2 0 2 0 9
4ENC_A 0.51 0.40 0.67 6 487 3 0 3 0 9

^top


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