Skip to main content

Table 3 Correlation measure performance of different NeEMO versions on the IM_631 dataset.

From: NeEMO: a method using residue interaction networks to improve prediction of protein stability upon mutation

  

r

ρ

τ

All

NeEMO

0.666

0.644

0.465

 

NeEMO_NOCC

0.637

0.626

0.447

 

NeEMO_NORING

0.618

0.603

0.430

Helix

NeEMO

0.645

0.612

0.436

 

NeEMO_NOCC

0.613

0.607

0.430

 

NeEMO_NORING

0.585

0.600

0.424

Beta

NeEMO

0.716

0.687

0.506

strand

NeEMO_NOCC

0.694

0.672

0.490

 

NeEMO_NORING

0.690

0.662

0.482

Coil

NeEMO

0.581

0.588

0.418

 

NeEMO_NOCC

0.546

0.560

0.391

 

NeEMO_NORING

0.502

0.501

0.350

Exposed

NeEMO

0.603

0.551

0.391

 

NeEMO_NOCC

0.553

0.516

0.360

 

NeEMO_NORING

0.522

0.498

0.350

Buried

NeEMO

0.638

0.614

0.441

 

NeEMO_NOCC

0.612

0.593

0.422

 

NeEMO_NORING

0.591

0.559

0.397

  1. NeEMO uses all input features, NeEMO_NOCC does not use node centralities, NeEMO_NORING does not use any RIN feature. Comparisons are shown for the entire dataset, on each of the 3 different secondary structure states and occurring in amino acids exposed to the solvent (e, RSA > 25%) or buried (b, RSA <= 25%).