Identification Of Age-correlated CircRNA Markers For The Development Of Forensic Age Estimation Models

Objective:Age estimation is one of the most important sections in forensic investigations.At present,the age of an unknown individual is mainly estimated through morphological indicators such as bones and teeth,which are subject to subjective factors and types of sample with low accuracy.What the toughest problem for current work is constructing high age prediction models by using biomarkers with better stability and availability.Recent studies have found that circular RNAs（circ RNAs）have long half-lives relative to m RNAs as they lack free ends and thus are not susceptible to degradation by exoribonucleases.They are found to upregulate globally during aging in multiple organisms.Thus,we propose that circ RNA can be used as a novel biomarker for individual age estimation.Methods:1. Selection of age-related circ RNAs through RNA-seq:here we collect peripheral whole blood from thirteen unrelated Chinese aged between 20 and62 years and analyze their RNA profiles.Three methods（bivariate correlation,lasso regression and support vector machine）are applied to select age-related circ RNAs candidates which will be validated further.2. Validation of circ RNA candidates using RT-q PCR:2ml blood samples from 50 unrelated volunteers aged between 19～72 years old are collected.circ RNA candidates’forward and reverse primers are designed for RT-q PCR testing.ΔCt-value(Ct _target–Ct _reference)represents circ RNAs’expression.3.Fitting individual age prediction models:50-sample expression data is obtained through RT-q PCR（80%are considered as training set and 20%are validation set）.Fitting methods include stepwise regression,lasso regression and random forest regression and 10-fold cross validation is used to select the best model.There are several parameters used for the assessment of a model,such as R square,a root-mean square error（RMSE）and mean absolute error（MAE）.Results:1.Here we identified more than 40,000 circ RNAs totally in blood of thirteen Chinese using RNA sequencing.Bivariate correlation analysis depicts that 197 circ RNAs as a whole have an absolute rho-value more than 0.6,among them,14 circ RNA candidates are selected by false discovery rate（FDR）adjusting with a q-value less than 0.01.Six circ RNAs are screened by lasso regression method and 9 by the feature selection of support vector machine（SVM）.These 28 circ RNAs are considered as candidates used for further analysis.2. All these 28 circ RNA candidates are designed divergent primers respectively.Primers are regarded as high-specificity primers when their PCR amplification products have a melting curve with one high peak in RT-q PCR and a single band in agarose gel electrophoresis.Additionally,their amplicons extracted from gels should pass Sanger sequence-consistency validation.3. According to every single circ RNA,we perform spearman correlation analysis which turns out 7 circ RNA candidates change with ageing among 50healthy unrelated individuals,one of them achieves the highest rho-value（hsa＿circ＿0000666,rho=-0.704,P<0.000）.Moreover,all these 7 circ RNAs change positively with age except hsa＿circ＿0000524.4. Three different modeling approaches including multiple stepwise regression,lasso regression and random forest regression are applied for fitting.R-square value for final three models are 0.69,0.84 and 0.87,RMSE are 10.1 years,8.0 years,5.7 years and MAE are 8.7 years,6.9 years,4.4 years,respectively.Conclusions:In current study,total RNA-seq profiles of thirteen biologically independent human peripheral whole blood specimens are performed.Analysis uncovers a strong bias for circ RNA upregulation during aging.7circ RNA candidates are thought to have significant relevance with age,and their changing patterns are identified in a 50-scale sample.Here we firstly use circ RNAs as biomarkers to construct age estimation models by introducing several statistical methods combined with machine learning algorithms.These prediction models demonstrate a relatively high accuracy.Thus,we propose that circ RNA can be used as a novel biomarker for forensic age estimation.