Skip to main content

Table 4 Comparison to existing methods

From: Improved accuracy of multiple ncRNA alignment by incorporating structural information into a MAFFT-based framework

   

Accuracy of predicted structure (MCC)

Method

Time (s.)

SPS

Pfold

McCaskill-MEA

RNAalifold

(intrinsic)

ClustalW (iterative)

98

0.669

0.488

0.554

0.482

 

ProbConsRNA

61

0.763

0.654

0.651

0.613

 

G-INS-i

12

0.768

0.622

0.646

0.622

 

LaRA 1.31

15,000

0.687

0.607

0.649

0.600

 

Murlet

64,000

0.773

0.712

0.702

0.668

 

MXSCARNA 2

700

0.769

0.718

0.712

0.666

 

RNA Sampler (fast)

19,000

0.641

0.659

0.684

0.662

0.655

RNA Sampler

70,000

0.655

0.685

0.703

0.705

0.705

MASTR

24,000

0.662

0.570

0.616

0.592

0.601

X-INS-i-larapair

15,000

0.758

0.665

0.692

0.672

 

X-INS-i-scarnapair

1,800

0.789

0.724

0.712

0.726

 
  1. The KKA dataset was used as the benchmark. The accuracies of alignments measured by the SPS criterion are listed in the SPS column. The SPS value was computed for each alignment and then averaged across all the alignments. The accuracies of predicted common secondary structures are shown in the four columns on the right. The alignment by each method was subjected to three external prediction programs, Pfold, McCaskill-MEA and RNAalifold, and then the differences from the Rfam curated structure were assessed. The MCC values were computed for each sequence and then averaged across all the sequences. The accuracy values for secondary structure internally predicted by RNA Sampler and MASTR are shown in the (intrinsic) column. The highest score in each column is underlined. The scores close to the highest (p > 0.01 in the Wilcoxon test) are shown in bold. McCaskill-MEA was run with the default value α = 0.91.