联合基因组-表型组指标的深蹲1 RM抗阻训练效果预测模型研究

A Combined Genome-Phenome Prediction Model for Squat 1 RM Resistance Training Effect

目的:应用全基因组关联分析(GWAS)筛选中国人深蹲1 RM抗阻训练效果遗传标记,构建基因组学预测模型;联合表型组学指标建立基因组-表型组训练效果预测综合模型,生物信息学分析其作用通路,以期为运动健身方案的精准化提供依据。方法:193名非规律运动成年人完成12周抗阻力量训练,干预前后测试深蹲1 RM等表型组学指标。应用Illumina Infinium CGA-24v1-0芯片进行全基因组测序,impute2填充基因型、plink进行全基因组关联分析(492万SNPs);应用向后回归去除冗余SNPs,向前回归建立深蹲1 RM训练效果基因组学预测模型;PRSice平均值法计算基于GWAS的权重后多基因得分(GPGS),线性回归建立GPGS与深蹲1 RM训练效果预测公式;逐步回归构建基因组学与表型组学指标联合的抗阻训练效果预测综合模型。HaploReg v4.1、GTEx、KEGG对纳入模型的SNPs进行生物功能注释。结果:(1)12周抗阻训练后,深蹲1 RM显著提高46%(P<0.01),个体干预效果范围为-20%~197%。(2)GWAS发现179个SNPs与深蹲1 RM抗阻训练效果显著关联(P<5×10^(-5)),10个SNPs被纳入基因组学预测模型(R2=0.496)。权重后的GPGS平均值为-0.44分(变化范围:-8.75~13.55);深蹲1 RM训练效果与GPGS呈正相关(R2=0.40,P<0.001);43.5%的受试者GPGS高于-0.44分,其抗阻训练效果高于平均值;联合基因组学与表型组学指标(深蹲初始值、躯干肌肉含量、全身脂肪含量)构建的综合模型R2=0.778。(3)生物信息学分析表明,深蹲1 RM训练效果基因组预测模型中的10个SNPs所在基因或受调控基因功能与骨骼肌生长发育、糖脂代谢调控等相关;受调控的基因富集于24条信号通路(P<0.05),经多重校验后,TGF-β信号通路、细胞周期、调节干细胞多能性等11条信号通路富集最为明显(FDR<0.05)。结论:首次基于GWAS构建的基因组学模型可有效预测深蹲1 RM训练效果,可解释个体差异的49.6%,GPGS高于-0.44分的受试者有更好的训练效果;联合基因组学-表型组学指标的综合模型预测能力提升至77.8%,为抗阻训练的精准化提供了可操作路径。TGF-β信号通路可能是预测模型中遗传标记对训练效果表型影响的关键信号通路。

Objective:Genome-wide association analysis(GWAS)was applied to screen genetic markers for resistance training effect of deep squat 1 RM in Chinese,and construct a genomic prediction model;a comprehensive genomic-phenomic training effect prediction model was established by combining phenomic indicators,and bioinformatics analysis of their action pathways,in order to provide a basis for the precision of exercise fitness programs.Methods:193 non-regularly exercising adults completed 12 weeks of resistance strength training,and phenomic indicators such as deep squat 1 RM were tested before and after the intervention.Illumina Infinium CGA-24v1-0 chip was applied for whole genome sequencing,impute2 populated genotypes,plink for genome-wide association analysis(4.92 million SNPs);backward regression was applied to remove redundant SNPs and forward regression was applied to build a genomic prediction model of the training effect of deep squat 1 RM;PRSice mean method was used to calculate GWAS-based weighted posterior polygenic score(GPGS),and linear regression was used to establish GPGS and deep squat 1 RM training effect prediction formula;stepwise regression was used to construct a comprehensive model for resistance training effect prediction with combined genomic and phenomic indicators.haploReg v4.1,GTEx and KEGG were for biofunctional annotation of SNPs incorporated into the model.Results:(1)after 12weeks of resistance training,deep squat 1 RM significantly improved by 46%(p<0.01),but individual intervention effects ranged from-20%to 197%;(2)GWAS found 179 SNPs significantly associated with deep squat 1 RM resistance training effects(p<5×10-5),and 10 SNPs were included in the genomics prediction model(R~2=0.496).The mean weighted GPGS score was-0.44(range of variation:-8.75 to 13.55);the effect of deep squat 1 RM training was positively correlated with GPGS(R~2=0.40,p<0.001);43.5%of subjects with GPGS above-0.44 had above-average resistance training effects;combined genomic and phenomic metrics(initial deep squat,trunk muscle content,whole-body fat content)to construct a comprehensive model R~2=0.778.(3)Bioinformatics analysis showed that the genes or regulated genes function of 10 SNPs in the genomic prediction model of deep squat 1 RM training effect were related to skeletal muscle growth and development and regulation of glucolipid metabolism;the regulated genes were enriched in 24 signaling pathways(p<0.05),and after multiple calibration of eleven signaling pathways,including TGF-βsignaling pathway,cell cycle,and signaling pathway regulating stem cell pluripotency,were most significantly enriched(FDR<0.05).Conclusions:the first GWAS-based genomic model can effectively predict the training effect of deep squat 1 RM,explaining 49.6%of individual differences,and subjects with GPGS scores higher than-0.44had better training effects;the predictive ability of the combined genomic-phenomic indexes of the integrated model was improved to 77.8%,providing an actionable path for the precision of resistance training.The TGF-βpathway may be a key signaling pathway for predicting the phenotypic effects of genetic markers in the model on training effects.