生物标志物
接收机工作特性
蛋白质组学
生物标志物发现
医学
生物信息学
计算生物学
内科学
化学
生物
生物化学
基因
作者
Zoltán Bán,Xuemin Gao,Xiaoming Ouyang,Jian-Pei Fang,Zixin Deng,Hao Wu,Yonghui Mao
标识
DOI:10.1101/2023.09.12.556305
摘要
ABSTRACT The development of blood-based multi-biomarker panels for screening diabetic patients, and as an easy-to-access tool for identifying individuals at greatest risk of developing diabetic kidney disease (DKD) and its progression, is essential. However, conventional blood biomarker-based methodologies (e.g. clinical tests and ELISA) are unable to predict DKD progression with high sensitivity and specificity. To overcome these challenges, we developed a deep, untargeted plasma proteome profiling technology (Proteonano™ platform) to identify potential multiple protein biomarkers involved in DKD progression. The Proteonano™ technology is an affinity selective mass spectrometric platform that comprises nanoparticle-based affinity binders (nanobinders) for low abundant protein enrichment, automated workflow for parallel sample preparation, and machine learning empowered bioinformatic software for data analysis. Using the Proteonano™ platform, we performed untargeted proteomics on 75 subjects (DKD progressors, n = 30; DKD non-progressors, n = 45) and identified an average of 953 ± 80 (AVG ± SD) protein groups, with a wide dynamic range of 8 orders of magnitude (with the lowest concentration down to 3.00 pg/mL). Among these, 38 proteins were differentially expressed between DKD progressors relative to non-progressors, and the predictive power for these proteins were assessed. Further, we performed random forest and LASSO analyses for additional variable selection. Variables selected by these approaches were assessed by Akaike information criterion method followed by ROC analysis, which identified a combination of multiple proteins (including VWF, PTGDS, B2M, BT3A2, and LCAT) that showed excellent predictive power over current methods, with an area under the curve value up to 0.97. Some of these plasma proteins are not previously recognized in the context of DKD progression, suggesting they are novel biomarkers. Our studies pave the way to develop multi-biomarker panels for DKD progression management. This study suggests that the Proteonano™ technology platform reported here can be employed as an established workflow enabling untargeted deep proteomic analysis to identify highly discriminative biomarkers for precise medicine.
科研通智能强力驱动
Strongly Powered by AbleSci AI