TY - JOUR
T1 - Exploring Prediction of Antimicrobial Resistance Based on Protein Solvent Accessibility Variation
AU - Marini, Simone
AU - Oliva, Marco
AU - Slizovskiy, Ilya B.
AU - Noyes, Noelle Robertson
AU - Boucher, Christina
AU - Prosperi, Mattia
N1 - Publisher Copyright:
© Copyright © 2021 Marini, Oliva, Slizovskiy, Noyes, Boucher and Prosperi.
PY - 2021/1/22
Y1 - 2021/1/22
N2 - Antimicrobial resistance (AMR) is a significant and growing public health threat. Sequencing of bacterial isolates is becoming more common, and therefore automatic identification of resistant bacterial strains is of pivotal importance for efficient, wide-spread AMR detection. To support this approach, several AMR databases and gene identification algorithms have been recently developed. A key problem in AMR detection, however, is the need for computational approaches detecting potential novel AMR genes or variants, which are not included in the reference databases. Toward this direction, here we study the relation between AMR and relative solvent accessibility (RSA) of protein variants from an in silico perspective. We show how known AMR protein variants tend to correspond to exposed residues, while on the contrary their susceptible counterparts tend to be buried. Based on these findings, we develop RSA-AMR, a novel relative solvent accessibility-based AMR scoring system. This scoring system can be applied to any protein variant to estimate its propensity of altering the relative solvent accessibility, and potentially conferring (or hindering) AMR. We show how RSA-AMR score can be integrated with existing AMR detection algorithms to expand their range of applicability into detecting potential novel AMR variants, and provide a ten-fold increase in Specificity. The two main limitations of RSA-AMR score is that it is designed on single point changes, and a limited number of variants was available for model learning.
AB - Antimicrobial resistance (AMR) is a significant and growing public health threat. Sequencing of bacterial isolates is becoming more common, and therefore automatic identification of resistant bacterial strains is of pivotal importance for efficient, wide-spread AMR detection. To support this approach, several AMR databases and gene identification algorithms have been recently developed. A key problem in AMR detection, however, is the need for computational approaches detecting potential novel AMR genes or variants, which are not included in the reference databases. Toward this direction, here we study the relation between AMR and relative solvent accessibility (RSA) of protein variants from an in silico perspective. We show how known AMR protein variants tend to correspond to exposed residues, while on the contrary their susceptible counterparts tend to be buried. Based on these findings, we develop RSA-AMR, a novel relative solvent accessibility-based AMR scoring system. This scoring system can be applied to any protein variant to estimate its propensity of altering the relative solvent accessibility, and potentially conferring (or hindering) AMR. We show how RSA-AMR score can be integrated with existing AMR detection algorithms to expand their range of applicability into detecting potential novel AMR variants, and provide a ten-fold increase in Specificity. The two main limitations of RSA-AMR score is that it is designed on single point changes, and a limited number of variants was available for model learning.
KW - AMR
KW - RSA
KW - antimicrobial resistance
KW - protein variant
KW - relative solvent accessibility
KW - scoring
KW - secondary structure
UR - http://www.scopus.com/inward/record.url?scp=85100511061&partnerID=8YFLogxK
UR - http://www.scopus.com/inward/citedby.url?scp=85100511061&partnerID=8YFLogxK
U2 - 10.3389/fgene.2021.564186
DO - 10.3389/fgene.2021.564186
M3 - Article
C2 - 33552147
AN - SCOPUS:85100511061
SN - 1664-8021
VL - 12
JO - Frontiers in Genetics
JF - Frontiers in Genetics
M1 - 564186
ER -