TMSNP: a web server to predict pathogenesis of missense mutations in the transmembrane region of membrane proteins

Abstract

The massive amount of data generated from genome sequencing brings tons of newly identified mutations, whose pathogenic/non-pathogenic effects need to be evaluated. This has given rise to several mutation predictor tools that, in general, do not consider the specificities of the various protein groups. We aimed to develop a predictor tool dedicated to membrane proteins, under the premise that their specific structural features and environment would give different responses to mutations compared to globular proteins. For this purpose, we created TMSNP, a database that currently contains information from 2624 pathogenic and 196 705 non-pathogenic reported mutations located in the transmembrane region of membrane proteins. By computing various conservation parameters on these mutations in combination with annotations, we trained a machine-learning model able to classify mutations as pathogenic or not. TMSNP (freely available at http://lmc.uab.es/tmsnp/) improves considerably the prediction power of commonly used mutation predictors trained with globular proteins.

MutTMPredictor: Robust and accurate cascade XGBoost classifier for prediction of mutations in transmembrane proteins

“…In this section, we compared MutTMPredictor with four predictors non-specific for membrane proteins (including fathmm [53] , PROVEAN [5] , SIFT [6] , and PolyPhen-2 [11] , [12] ) and two predictors specific for membrane proteins (i.e. Pred-MutHTP [31] and TMSNP [35] ). The performance results are documented in Table 14 , Table 15 .…”

“…Accordingly, we removed 13 disease-associated and 116 neutral mutations. (4) TMSNP database: TMSNP [35] stored a certain set of membrane proteins taken from UniProt [45] . It retrieved the disease-causing/pathogenic mutations occurring in transmembrane helical regions from ClinVar [40] and SwissVar [36] and nonpathogenic mutations/allele frequency from GnomAD [46] and ClinVar [40] .…”

“…Based on the confusion matrix, several performance metrics can be derived, such as Recall , sensitivity ( Sen ), specificity ( Spe ), precision ( Pre ), accuracy ( ACC ), Matthew's Correlation Coefficient ( MCC ), F 1 -score ( F 1 ), and negative predictive value ( NPV ), error rate ( ER) , false negative rate ( FNR ), and false positive rate ( FPR ). In this work, the above performance metrics were calculated to evaluate the developed predictor and other existing predictors [31] , [34] , [35] . where TP (true positive)/ TN (true negative) are the number of disease-associated mutations/neutral mutations that are correctly predicted as disease-associated/neutral mutations; FP (false positive) is the number of neutral mutations that are incorrectly predicted as disease-associated mutations; FN (false negative) is the number of disease-associated mutations that are incorrectly predicted as neutral mutations, respectively.…”

“…The MutHTP (mutations in human transmembrane proteins) database were developed to deposit and retrieve mutations in such proteins [32] . Besides, BorodaTM [33] , Pred-MutHTP [31] , mCSM-membrane [34] , and TMSNP [35] are specific predictors for mutations prediction in such proteins. Specifically, in 2019, Kulandaisamy et al collected MM in transmembrane proteins from HumSavar ( http://www.uniprot.org/docs/humsavar ), SwissVar [36] , 1000 Genomes [37] , ExAC [38] , COSMIC [39] , and ClinVar [40] .…”

“…However, mCSM-membrane could only predict mutations in proteins with known 3D structures in PDB database [41] . In 2021, TMSNP database was constructed, which comprised 196,705 non-pathogenic, 2,624 pathogenic, and 437 likely pathogenic mutations, respectively in transmembrane proteins [35] .…”