Prediction of transporter family from protein sequence by support vector machine approach

Abstract

Transporters play key roles in cellular transport and metabolic processes, and in facilitating drug delivery and excretion. These proteins are classified into families based on the transporter classification (TC) system. Determination of the TC family of transporters facilitates the study of their cellular and pharmacological functions. Methods for predicting TC family without sequence alignments or clustering are particularly useful for studying novel transporters whose function cannot be determined by sequence similarity. This work explores the use of a machine learning method, support vector machines (SVMs), for predicting the family of transporters from their sequence without the use of sequence similarity. A total of 10,636 transporters in 13 TC subclasses, 1914 transporters in eight TC families, and 168,341 nontransporter proteins are used to train and test the SVM prediction system. Testing results by using a separate set of 4351 transporters and 83,151 nontransporter proteins show that the overall accuracy for predicting members of these TC subclasses and families is 83.4% and 88.0%, respectively, and that of nonmembers is 99.3% and 96.6%, respectively. The accuracies for predicting members and nonmembers of individual TC subclasses are in the range of 70.7-96.1% and 97.6-99.9%, respectively, and those of individual TC families are in the range of 60.6-97.1% and 91.5-99.4%, respectively. A further test by using 26,139 transmembrane proteins outside each of the 13 TC subclasses shows that 90.4-99.6% of these are correctly predicted. Our study suggests that the SVM is potentially useful for facilitating functional study of transporters irrespective of sequence similarity.

A support vector machines approach for virtual screening of active compounds of single and multiple mechanisms from large libraries at an improved hit-rate and enrichment factor

“…Statistically, a substantial percentage of active members can be classified by ML methods as active even if its family representative is in the inactive class [57]. Therefore, in principle, a reasonably good ML model can be derived from these putative inactive samples, which has been confirmed by a number of studies [53][54][55]57].…”

“…Apart from the use of known inactive compounds and active compounds of other biological target classes as putative inactive compounds [7][8][9][11][12][13][14]40], a new approach extensively used for generating inactive proteins in ML classification of various classes of proteins [53][54][55] may be applied for generating putative inactive compounds. An advantage of this approach is its independence on the knowledge of known inactive compounds and active compounds of other biological target classes, which enables more expanded coverage of the ''inactive'' chemical space in cases of limited knowledge of inactive compounds and compounds of other biological classes.…”

Prediction of antibiotic resistance proteins from sequence-derived properties irrespective of sequence similarity

“…This may likely result from the availability of a more diverse set of negative data than that for the positive data, which enables the SVM to perform a better statistical learning for recognition of non-ARPs. Moreover, a SVM based on an imbalanced data set tends to produce feature vectors that push the hyperplane towards the side with a smaller number of data, which can lead to a reduced accuracy for the data set either with a smaller number of samples or of less diversity [30]. This may be another reason why the prediction accuracy for ARPs is generally lower than that of non-ARPs.…”

“…The negative data set, representing non-ARPs, was selected by a commonly used procedure [29,30]. In this procedure, representative proteins of curated protein families in the Pfam database [31] that contain no single known ARPs are selected as non-ARPs.…”

Support vector machines coupled with proteomics approaches for detecting biomarkers predicting chemotherapy resistance in small cell lung cancer

“…SVM is a new machine learning approach originally proposed and developed by Vapnik (18). SVM applications are being actively pursued in various areas recently, from face recognition to genomics (19). It is a powerful tool for analyzing complex data derived from SELDI-TOF MS. We constructed a non-linear SVM classifier with a radial based function (RBF) kernel, and with the parameter Gamma 0.6, being the cost of the constrain violation 19 to discriminate the different groups.…”