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 PPfold(20) - scored higher in this pairwise comparison

  4. Performance of ProbKnot - scored lower in this pairwise comparison

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


Overview

Metric PPfold(20) ProbKnot
MCC 0.691 > 0.565
Average MCC ± 95% Confidence Intervals 0.679 ± 0.109 > 0.582 ± 0.115
Sensitivity 0.544 > 0.495
Positive Predictive Value 0.881 > 0.650
Total TP 288 > 262
Total TN 74040 > 73964
Total FP 49 < 163
Total FP CONTRA 0 < 20
Total FP INCONS 39 < 121
Total FP COMP 10 < 22
Total FN 241 < 267
P-value 2.59578924681e-08

^top




Performance plots


  1. Comparison of performance of PPfold(20) and ProbKnot. 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 PPfold(20) and ProbKnot).

  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 PPfold(20) and ProbKnot).

  3. Comparison of average Matthews Correlation Coefficients (MCCs) for PPfold(20) and ProbKnot. 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 PPfold(20) and ProbKnot).

^top





Performance of PPfold(20) - scored higher in this pairwise comparison

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

Total Base Pair Counts
Total TP 288
Total TN 74040
Total FP 49
Total FP CONTRA 0
Total FP INCONS 39
Total FP COMP 10
Total FN 241
Total Scores
MCC 0.691
Average MCC ± 95% Confidence Intervals 0.679 ± 0.109
Sensitivity 0.544
Positive Predictive Value 0.881
Nr of predictions 15

^top



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

RNA Chain Rfam family MCC SENS PPV TP TN FP FP CONTRA FP INCONS FP COMP FN
2L94_A 0.63 0.45 0.90 9 980 1 0 1 0 11
3AMU_B 0.86 0.74 1.00 20 2983 1 0 0 1 7
3J20_1 0.96 0.91 1.00 21 2905 0 0 0 0 2
3J20_0 0.84 0.70 1.00 21 2829 0 0 0 0 9
3J2L_3 0.78 0.62 0.97 33 7841 3 0 1 2 20
3RKF_A 0.73 0.56 0.95 19 2191 1 0 1 0 15
3SD1_A 0.68 0.52 0.88 22 3891 3 0 3 0 20
3ZEX_C 0.33 0.21 0.52 11 14175 12 0 10 2 41
3ZEX_D 0.81 0.71 0.92 35 6983 3 0 3 0 14
4A1C_3 0.77 0.63 0.94 34 7104 2 0 2 0 20
4A1C_2 0.21 0.15 0.29 5 11764 16 0 12 4 28
4AOB_A 0.74 0.60 0.93 25 4344 3 0 2 1 17
4ENB_A 0.56 0.37 0.88 7 1267 1 0 1 0 12
4ENC_A 0.58 0.42 0.80 8 1316 2 0 2 0 11
4FRG_B 0.73 0.56 0.95 18 3467 1 0 1 0 14

^top



Performance of ProbKnot - scored lower in this pairwise comparison

1. Total counts & total scores for ProbKnot

Total Base Pair Counts
Total TP 262
Total TN 73964
Total FP 163
Total FP CONTRA 20
Total FP INCONS 121
Total FP COMP 22
Total FN 267
Total Scores
MCC 0.565
Average MCC ± 95% Confidence Intervals 0.582 ± 0.115
Sensitivity 0.495
Positive Predictive Value 0.650
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
2L94_A 0.97 0.95 1.00 19 971 0 0 0 0 1
3AMU_B 0.65 0.59 0.73 16 2981 8 0 6 2 11
3J20_1 0.71 0.70 0.73 16 2904 7 0 6 1 7
3J20_0 0.45 0.40 0.52 12 2827 12 0 11 1 18
3J2L_3 0.60 0.49 0.74 26 7840 11 0 9 2 27
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
3ZEX_C 0.38 0.31 0.47 16 14162 21 2 16 3 36
3ZEX_D 0.77 0.67 0.89 33 6984 4 0 4 0 16
4A1C_3 0.73 0.61 0.87 33 7102 6 1 4 1 21
4A1C_2 0.13 0.15 0.12 5 11738 49 7 31 11 28
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
4FRG_B 0.37 0.31 0.45 10 3464 12 3 9 0 22

^top


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