Protein intrinsically disordered region prediction by combining neural architecture search and multi-objective genetic algorithm

Table 3 Performance of different methods on Disprot and Disprot-PDB in the CAID1

Predictor	AUC^a	Rank^b	AUC^c	Rank^d	RS^e
IDP-Fusion	0.802	2	0.925	1	3
DeepIDP-2L [18]	0.796	3	0.918	3	6
SPOT-Disorder2 [14]	0.76	7	0.920	2	9
fIDPnn [34]	0.814	1	0.873	9	10
RawMSA [35]	0.78	4	0.894	6	10
SPOT-Disorder [7]	0.757	8	0.916	4	12
AUCpreD [8]	0.757	8	0.906	5	13
ESPritz-D [36]	0.774	5	NA	NA	16
DisoMine [37]	0.765	6	NA	NA	17
Predisorder [38]	0.747	10	0.878	7	17
DISOPRED3 [6]	NA	NA	0.875	8	19
IsUnstruct [39]	NA	NA	0.868	10	21

^aRepresents the AUC value obtained by various methods on Disprot
^bRepresents the ranking of the AUC value obtained by various methods on Disprot
^cRepresents the AUC value obtained by various methods on Disprot-pdb
^dRepresents the ranking of the AUC value obtained by various methods on Disprot-pdb
^eRepresents Ranking Score (RS), which is the sum of rank^a and rank^b. The smaller the value is, the better the performance of the corresponding method is. If the values of rank^a or rank^b are not available, it indicates that the corresponding method is not the top 10 best method in the CAID1. Therefore, its rank^a or rank^b is set as 11

ISSN: 1741-7007