Multidrug Machine Learning Models for Predicting Human Immunodeficiency Virus‐1 Reverse Transcriptase Inhibitor Resistance

Drug resistance remains a major barrier to HIV/AIDS treatment, underscoring the need for rapid evaluation of resistance mechanisms and candidate inhibitors. We present a multidrug machine‐learning framework using drug‐isolate fold‐change (DIF) data to predict resistance to human immunodeficiency virus type 1 (HIV‐1) reverse transcriptase inhibitors (RTIs). Unlike conventional isolate fold‐change (IF) models, DIF integrates viral mutation profiles with inhibitor‐specific molecular information, enabling simultaneous assessment of multiple RTIs, including unseen compounds. Artificial neural networks (ANN), XGBoost (XGB), LightGBM (LGBM), random forest (RF), and support vector regression (SVR) were evaluated. Across all algorithms, class‐specific DIF models consistently yielded higher correlation coefficients and lower mean squared errors than corresponding IF models. XGB‐DIF performed best overall, achieving a mean Pearson correlation coefficient of 0.914 and a mean squared error of 0.064. The models also captured pairwise resistance trends among FDA‐approved RTIs. External validation against an independent ChEMBL dataset yielded an accuracy of 0.824, an AUC of 0.901, and a Pearson correlation coefficient of 0.773 for the best‐performing XGB model. For clinically relevant variants mean absolute errors in logarithmic fold change were as low as 0.057, with near‐perfect resistance ranking for K103N, Y181C, and Y188L. These findings support the robustness and potential utility of the DIF framework.

Authors

Institutions

Publication Details

Journal
ChemMedChem
Published
2026-11-16
DOI
https://doi.org/10.1002/cmdc.70556
Citations
2
Primary Topic
Computational Drug Discovery Methods
Type
article
Field-Weighted Citation Impact
1.32

Funders

Controls
|||
ALL TIME
JAN
FEB
MAR
APR
MAY
JUN
JUL
AUG
SEP
OCT
NOV
article

Multidrug Machine Learning Models for Predicting Human Immunodeficiency Virus‐1 Reverse Transcriptase Inhibitor Resistance

Murat Sarı, Hüseyin Tunç, Sümeyye Yılmaz
2 citations
ChemMedChem
Computational Drug Discovery Methods
1.32
article

Multidrug Machine Learning Models for Predicting Human Immunodeficiency Virus‐1 Reverse Transcriptase Inhibitor Resistance

Murat Sarı, Hüseyin Tunç, Sümeyye Yılmaz
article en
2 citations

Abstract

Drug resistance remains a major barrier to HIV/AIDS treatment, underscoring the need for rapid evaluation of resistance mechanisms and candidate inhibitors. We present a multidrug machine‐learning framework using drug‐isolate fold‐change (DIF) data to predict resistance to human immunodeficiency virus type 1 (HIV‐1) reverse transcriptase inhibitors (RTIs). Unlike conventional isolate fold‐change (IF) models, DIF integrates viral mutation profiles with inhibitor‐specific molecular information, enabling simultaneous assessment of multiple RTIs, including unseen compounds. Artificial neural networks (ANN), XGBoost (XGB), LightGBM (LGBM), random forest (RF), and support vector regression (SVR) were evaluated. Across all algorithms, class‐specific DIF models consistently yielded higher correlation coefficients and lower mean squared errors than corresponding IF models. XGB‐DIF performed best overall, achieving a mean Pearson correlation coefficient of 0.914 and a mean squared error of 0.064. The models also captured pairwise resistance trends among FDA‐approved RTIs. External validation against an independent ChEMBL dataset yielded an accuracy of 0.824, an AUC of 0.901, and a Pearson correlation coefficient of 0.773 for the best‐performing XGB model. For clinically relevant variants mean absolute errors in logarithmic fold change were as low as 0.057, with near‐perfect resistance ranking for K103N, Y181C, and Y188L. These findings support the robustness and potential utility of the DIF framework.

ChemMedChemVol. 21(22)
Bahçeşehir University (TR), Biruni University (TR), Istanbul Technical University (TR)
Türkiye Bilimsel ve Teknolojik Araştırma Kurumu
Good health and well-being
Openalex Percentile: Top 24%
Computational Drug Discovery Methods
1.32
AI Navigator

Ask Laika to Summarize, Analyze, and Connect papers live on the map.

Summarize Papers & Methodologies

Extract key findings, datasets, and comparative methods across publications.

Benchmark Rankings & Visual Analytics

Rank top research institutions, authors, funders, topics, and journals by Field-Weighted Citation Impact (FWCI) and paper volume with instant charts.

Connect Distant Disciplines

Bridge topological clusters on the map to find hidden collaborative intersections.

Laika access coming soon