Ensemble-PlantSCL: Ensemble of multiple classifiers for multilabel classification of plant protein subcellular localization

Abstract: The accurate prediction of protein localization is a critical step in any functional genome annotation process. This paper proposes an improved strategy for protein subcellular localization prediction in plants based on multiple classifiers to improve prediction results in terms of both accuracy and reliability. The prediction of plant protein subcellular localization is challenging because the underlying problem is not only a multiclass, but also a multilabel problem. Generally, plant proteins can be found in 10-14 locations/compartments. The number of proteins in some compartments (nucleus, cytoplasm, and mitochondria) is generally much greater than that in other compartments (vacuole, peroxisome, Golgi, and cell wall). Therefore, the problem of imbalanced data usually arises. Therefore, we propose an ensemble machine learning method based on average voting among heterogeneous classifiers. We first extracted various types of features suitable for each type of protein localization to form a total of 479 feature spaces. Then, feature selection methods were used to reduce the dimensions of the features into smaller informative feature subsets. This reduced feature subset was then used to train/build 3 different individual models. In the process of combining the 3 distinct classifier models, we used an average voting approach to combine the results of these 3 different classifiers that we constructed to return the final probability prediction. The method can predict subcellular localizations in both single- and multilabel locations based on the voting probability. Experimental results indicated that the proposed ensemble method can achieve correct classification with an overall accuracy of 84.58% for 11 compartments on the basis of the testing dataset.

Please cite:

Wattanapornprom, W., Thammarongtham, C., Hongsthong, A., Lertampaiporn, S. Ensemble of multiple classifiers for multilabel classification of plant protein subcellular localization.2021. Life 11 (4), 293.

Download Link: Training and Testing data, Standalone Program

Command: perl EnsemblePlantSCL.pl input_fasta_file (mode1: probability for predicted location only)

perl EnsemblePlantSCL_distribution.pl input_fasta_file (mode2: probability distribution for all locations)

***** Fasta seq must has length > 10 aa and contains only standard amino acids *****

###Example:

perl EnsemblePlantSCL_distribution.pl ./Seq_Data/pln_tag_EMR.fa

###Output:

############ Prediction results by EnsemblePlantSCL (mode2) ############

(Probability distribution of 1:Cell membrane,2:Cell wall,3:Cytoplasm,4:ER,5:Extra,6:Golgi,7:mitochondria,8:Nucleus,9:Peroxisome,10:Plastid,11:Vacuole)

Seq.1 predicted as 4:ER - with prob. dist: 0.035,0.005,0.253,*0.588,0.007,0.007,0.043,0.01,0.001,0.044,0.007

Seq.2 predicted as 4:ER - with prob. dist: 0.032,0,0.161,*0.401,0.004,0.012,0.214,0.144,0.002,0.02,0.01

Seq.3 predicted as 4:ER - with prob. dist: 0.041,0.001,0.344,*0.45,0.028,0.003,0.042,0.05,0.003,0.031,0.007

Seq.4 predicted as 4:ER - with prob. dist: 0.067,0.002,0.131,*0.4,0.015,0.035,0.203,0.083,0.004,0.042,0.018

Seq.5 predicted as 4:ER - with prob. dist: 0.036,0,0.142,*0.408,0.003,0.031,0.314,0.059,0,0.004,0.002