CN114045333B - Method for predicting age by pyrosequencing and random forest regression analysis - Google Patents

Abstract

The present invention provides a method for age prediction. The method includes pyrosequencing and random forest regression analysis. The random forest regression analysis model is constructed using R package random Forest, and determined by forward selection method. optimal site combination. The method provided by the invention only needs 0.1 ng of template DNA, which can be used for forensic bloodstain samples with relatively high difficulty; the whole process can be completed within 10 hours; according to gender differences, two independent age prediction models are established by gender; only 3 ‑4 CpG loci, the accuracy of predicting age can reach MAD<3 years.

Description

Methods for Age Prediction Using Pyrosequencing and Random Forest Regression Analysis

technical field

The invention belongs to the field of forensic medicine, and in particular relates to a method for age prediction using pyrosequencing and random forest regression analysis.

Background technique

The biological age assessment of donors of unknown samples is one of the most important tools in forensic investigations. It narrows down the criminal suspects, which in turn complements the offender's externally visible feature prediction and biogeographic ancestry inference. Previously established age classification methods involve morphological analysis of skeletal features. When solid tissues such as bones and teeth are available, age can be accurately determined by anthropological methods. However, it is difficult to use such methods in practice as other tissues, such as bodily fluids, are more likely to be encountered during forensic investigations. Recently, several molecular-level-based methods have been proposed to estimate age, including telomere length analysis, age-dependent deletions of mitochondrial DNA or T-cell DNA rearrangements, and protein alterations such as racemization of aspartate and Advanced glycation end products. However, all these methods have limitations that limit their applicability to crime scenes, especially their low accuracy and strict sample requirements. For example, the standard error of age prediction based on the quantification of signaling combined with T-cell receptor rearrangement excision circles (sjTRECs) is ±8.0 years.

A possible alternative to these methods is the detection of epigenetic modifications, such as methylation, which are now known to vary with age. To date, studies on forensic age prediction have mainly focused on whole blood samples, with mean absolute deviation (MAD) ranging from 3 to 10 years, mainly using multiple linear regression models. A small number of studies have achieved relatively low prediction errors (3.24-4.7 years) using machine learning algorithms such as Support Vector Machines (SVM), Artificial Neural Networks (ANN), and Random Forest Regression (RFR); however, these studies only in fresh body fluids. In addition, age prediction based on scarring (more common in crime scene investigations) has not been systematically studied.

Therefore, the present invention aims to establish a sensitive, fast, and reliable method for age prediction based on pyrosequencing technology and random forest regression, which is suitable for age prediction of various samples including bloodstains.

SUMMARY OF THE INVENTION

The present invention screens out a set of DNA methylation age prediction sites for analyzing the samples in forensic cases from the genome sequence, designs primers for each site, and uses pyrosequencing technology to analyze the methylation level of each site. Analysis, and then use random forest regression to build age prediction models for males and females separately. The aim is to invent a sensitive, fast, reliable and high-accuracy age prediction analysis method using fewer sites. This detection method can be used for age prediction of blood stains and other samples. Primer design and sequencing protocols were optimized.

the term:

RFR: Random Forest Regressor, random forest regression.

SVR: Support Vector Regression, support vector regression.

MAD: Mean Absolute Deviation, mean absolute error.

In one aspect, the present invention provides a method for age prediction.

The method includes pyrosequencing and random forest regression analysis. The random forest regression analysis model is constructed using the R package random Forest, and the forward selection method is used to determine the best site combination.

The parameter settings in the random forest regression analysis model construction in the described method: the mtry parameter is the same as the number of CpG sites for each modeling, the minimum node size is 5, and the number of trees is set to 1000.

The random forest regression analysis model selected age-related DNA methylation markers located in ELOVL2, C1orf132, TRIM59, KLF14, FHL2 and NPTX2 genes, respectively.

When the described random forest regression established the age prediction model, a total of 7 age-related DNA methylation sites were selected, of which 3 were male: TRIM59.pos7, KLF14.pos2, ELOVL2.pos7; 4 females were TRIM59 .pos8, KLF14.pos3, Clorf132.pos2 and FHL2.pos6.

The volume of PCR product used in the described pyrosequencing was 12 μL.

In some embodiments, the method includes the following steps:

(1) DNA extraction;

(2) sulfite conversion;

(3) PCR;

(4) Pyrosequencing;

(5) Model prediction.

In another aspect, the present invention provides a set of gene combinations for age prediction by random forest regression analysis.

The gene panel includes ELOVL2, C1orf132, TRIM59, KLF14, FHL2 and NPTX2.

The methylation sites include male-related sites: TRIM59.pos7, KLF14.pos2, ELOVL2.pos7 and female-related sites: TRIM59.pos8, KLF14.pos3, Clorf132.pos2, FHL2.pos6.

In yet another aspect, the present invention provides a set of primers for age prediction by random forest regression analysis.

The primers described were used for pyrosequencing.

The primers and their sequencing sites are as follows:

The primer sequences F, R, and S represent forward primer, reverse primer and sequencing primer, respectively, and the label biotin before the sequence indicates that the primer is labeled with biotin.

In yet another aspect, the present invention provides the aforementioned method and/or application of gene combination and/or methylation site and/or primer in preparing a kit for predicting age.

In yet another aspect, the present invention provides a kit for predicting age.

The kit includes the following primers:

The kit also includes other reagents for pyrosequencing.

The kit was used in conjunction with a random forest regression model.

Beneficial effects of the present invention:

(1) Only 0.1ng of template DNA is needed, which can be used for difficult forensic blood stains

Many techniques, such as EpiTYPER, Snapshots, pyrosequencing, and massively parallel sequencing (MPS), can provide relatively accurate DNA methylation measurements. One of the main reasons that limit the application of EpiTYPER analysis in forensics is that it requires up to 1 μg of genomic DNA. However, it is difficult to obtain such a high amount of DNA in actual crime scene investigations, and bodily fluid stains are often encountered at crime scenes. Compared with EpiTYPER, the template DNA required for MPS can be reduced to 10ng, and Snapshots require 4ng template DNA. In the present invention, however, accurate age prediction can be performed using 0.1 ng of template DNA. Previous studies have shown that 10-20ng of template DNA is required for successful age prediction based on methylation. Therefore, the detection method of the present invention has the highest sensitivity in the existing blood stain detection, and has a good forensic application prospect.

(2) The whole process can be completed within 10 hours

The method of the present invention can be completed in one day, much faster than other available methods. DNA extraction/quantification, sodium bisulfate conversion, PCR and pyrosequencing assays took 2h, 2.5h, 3h and 2h, respectively. In comparison, the standard procedures for both EpiTyper and MPS take more than 2 days. In particular, MPS requires specialized equipment and a complex bioinformatics analysis system, which is difficult to complete within 3 days.

(3) For gender differences, establish two independent age prediction models by gender

Select random forest regression (RFR) to establish age prediction model, using 3 males (TRIM59.pos7, KLF14.pos2, ELOVL2.pos7) and 4 females (TRIM59.pos8, KLF14.pos3, Clorf132.pos2) and FHL2.pos6) loci, the final model for a total of 7 loci had a mean absolute error (MAD) of prediction for males and females of 2.8 years (R=0.99) and 2.93 years (R=0.98), respectively.

(4) Using only 3-4 CpG sites, the accuracy of predicting age can reach MAD<3 years

A meta-analysis of age-prediction studies over the past few years showed that, in age-prediction models established by previous studies, nearly all MADs were >3 years old. Our model is the most efficient due to the use of RFR (MAD < 3 years and requires only 3-4 CpG loci, using only 3 CpG loci in male samples and 4 CpG loci in female samples). The age prediction model of the present invention, which has few sites and still maintains high accuracy, is more practical for forensic inference.

Description of drawings

Figure 1 shows that random forest regression (RFR) outperforms support vector regression (SVR) in age prediction.

Figure 2 shows the predicted age versus actual age for the random forest regression (RFR) test dataset.

Figure 3 shows the sensitivity detection of seven methylation markers in trace amounts of DNA.

Figure 4 shows the correlation analysis between the methylation levels of 7 CpG sites and age.

Figure 5 is a comparison of the accuracy of age prediction methods with published studies.

Detailed ways

The present invention will be described in further detail below with reference to specific embodiments. The following embodiments are not intended to limit the present invention, but are only used to illustrate the present invention. The experimental methods used in the following examples, unless otherwise specified, the experimental methods that do not specify specific conditions in the examples are usually in accordance with conventional conditions, and the materials, reagents, etc. used in the following examples, unless otherwise specified, are all Commercially available.

Example 1 DNA extraction and site screening

(1) DNA extraction:

Optimize the DNA extraction protocol to reduce the loss of trace DNA in blood.

The accuracy of methylation analysis depends on extracting high-quality DNA from bloodstains. The QIAamp DNAInvestigator kit has been recognized as a more reliable method for DNA extraction from forensic samples, resulting in successful high-quality DNA extraction within 2 hours. We have further optimized the kit to include shorter incubation times at higher temperatures, addition of carrier RNA to the lysate, and heating reagents to dissolve the DNA.

The previous method was to incubate the sample at 56°C for 1 hour, and the improved method is to incubate the sample at 85°C for 10 minutes, followed by a second incubation at 56°C for 1 hour to increase the amount of bloodstained DNA extracted.

Heating reagents that lyse DNA can also accelerate cell shedding on blood spots and increase DNA lysis, thereby reducing the loss of trace amounts of DNA.

(2) Site selection:

Based on the literature, six age-related DNA methylation marks, located at ELOVL2, C1orf132, TRIM59, KLF14, FHL2, and NPTX2, were selected to ensure that we focused on age-related regions.

(3) Primer design:

Since the DNA obtained from bloodstains is extremely small in quantity and of low quality, the accuracy and sensitivity of PCR is critical. PCR primers and sequencing primers were designed using PyroMark Assay Design version 2.0 (Qiagen, Germany). When designing primers, adjust the target sequence so that the primers contain as many cytosines (C) as possible to detect more methylation sites. We avoided SNPs and other polymorphisms in the target region because they could bias the sequencing reaction. Furthermore, possible methylation sites in the primer-binding sequence were excluded, the GC content was kept below 60%, and primers with high specificity (ie, no primer-dimer formation) were selected. When necessary, we modified published methods (eg, adding dimethyl sulfoxide (DMSO) to avoid dimer formation) to optimize the protocol. Among the PCR primers, the 5' end of one primer should be labeled with biotin to bind to streptavidin-coated magnetic beads for subsequent separation and purification of single-stranded PCR products, and the other should not be labeled. Biotin-labeled primers contain free biotin. Free biotin will compete with the template for binding to streptavidin-coated magnetic beads and reduce the signal level. HPLC-purified biotin-labeled primers must be used. The amplicon length of each gene of interest ranged from 105-306 bp. The final primers are shown in the table below:

Table 1 PCR primers, pyrosequencing primers and CpG sequences for age-related methylation analysis

Example 2 Detection of DNA methylation by pyrosequencing technology

(1) bisulfite conversion

Extracted DNA (40 μL) was subjected to bisulfite conversion using EpiTect fast DNA bisulfite kit (Qiagen, Germany). DNA samples were mixed with CT conversion reagent (bisulfite kit) to obtain a final volume of 140 μL of product, then incubated at 95°C for 5 minutes, 60°C for 20 minutes, and then purified.

(2) PCR

The reaction mixture (25 μL) contained 2 μL of transforming DNA, 12.5 μL PCR master mix (Qiagen, Germany) and 0.1-0.5 mM primers. Primer concentrations were adjusted to obtain dimer-free specific DNA products. Thermal cycling conditions were as follows: denaturation at 95°C for 10 min; 45 cycles at 95°C for 30 s, 30 s at 56°C (NPTX2 58°C, 30 s), 30 s at 72°C; A final extension was performed at 72°C for 5 minutes. Electrophoretic detection was performed using agarose gel electrophoresis.

(3) Pyrosequencing

Templates prepared from biotin-labeled PCR amplification products were sequenced using a Pyromark Q48 thermal sequencer (Qiagen, Germany) and a Pyro-Gold kit (Qiagen, Germany). During the previous pyrosequencing procedure, the volume of PCR product was 10 μL, which produced an unstable signal that was indistinguishable from the background signal. Our method increases the volume of PCR products to 12 μL, which can effectively avoid the generation of unstable signals.

Example 3 Construction of bloodstain age prediction model

(1) Compare the age prediction accuracy of SVR and RFR models

Our previous findings showed that the SVR model was more accurate than methods such as multiple linear regression, multiple nonlinear regression, and backpropagation neural networks, therefore, we utilized the SVR and RFR models, based on the combination of all 46 CpG sites, to establish the most The best fit age prediction model is calculated and its prediction accuracy is calculated. The SVR model was built in R package e1071, with parameter settings: cost=2, gamma=0.8, epsilon=0.1. The RFR model was constructed with the R package random Forest, the mtry parameter was the same as the number of CpG sites for each modeling, the minimum node size was 5, and the number of trees was set to 1000.

In order to improve the calculation speed, the forward selection method was used to determine the optimal site combination. Whole blood samples from 241 bloodstain samples (age range 10-79 years) from 241 healthy Chinese Han volunteers, including 128 males and 113 females. All donors provided informed consent, Chinese Academy of Sciences Beijing Genome The Institute passed the ethical approval of this study) randomly selected 70% of the samples to form the training dataset and the remaining 30% as the test dataset to evaluate the accuracy of the RFR model. The training was repeated 100 times, each time selecting the best site (ie, the smallest MAD). The site with the highest recording frequency was selected as a suitable site for the final model. In the two-site training model, after the best site, the site with the highest frequency and the smallest MAD was recorded as the second best site.

The age prediction model constructed by RFR, females use 4 loci (TRIM59.pos8, KLF14.pos3, Clorf132.pos2 and FHL2.pos6) males use 3 loci (TRIM59.pos7, KLF14.pos2, ELOVL2.pos7), Resulting MADs < 3 years. Under the SVR model, even though both males and females had 8 loci, the MAD was stable at around 4.5 years, a result suggesting that RFR outperformed SVR in predicting age (Fig. 1).

(2) Test data set to verify the prediction accuracy

The remaining 30% of the blood stain samples (38 males and 33 females) were used as the test dataset to verify the 7 loci (3 loci in males: TRIM59.pos7, KLF14.pos2, ELOVL2.pos7; female 4 loci: TRIM59.pos8, KLF14.pos3, Clorf132.pos2, and FHL2.pos6) age prediction accuracy, resulting in a predicted MAD of 2.8 years for males and females (R=0.99) and 2.93 years (R=0.98) (Figure 2).

Example 4 Sensitivity detection

Whole blood samples were collected from 241 healthy Chinese Han volunteers (128 males and 113 females) ranging in age from 10-79 years. All donors provided informed consent, and the Beijing Institute of Genomics, Chinese Academy of Sciences received ethical approval for this study.

Bloodstains were prepared by aliquoting 20 μL of whole blood onto filter paper and then stored at room temperature for 1 year. To determine detection sensitivity, DNA extracted from bloodstains was serially diluted to 100, 50, 10, 5, 2.5, 1.0, 0.50, 0.25, and 0.10 ng. Bloodstain samples with different concentrations were all subjected to methylation analysis, firstly bisulfite conversion, and then PCR amplification and pyrosequencing (refer to the method of Example 1). The difference in methylation percentage between 0.1 ng DNA and higher DNA concentrations was compared to judge the sensitivity of our proposed methylation detection method in blood stain detection.

We observed 4 CpG loci (TRIM59.pos8, KLF14.pos3, Clorf132.pos2 and FHL2.pos6) in females and 3 CpG loci in males (TRIM59.pos7, KLF14.pos2, ELOVL2.pos7) for age prediction ), there was no significant difference in percent methylation between 0.1 ng of DNA and higher concentrations of DNA (P≥0.05, KS test; Figure 3). For the ELOVL2.pos7 locus, 1.0 ng of DNA is required to achieve similar levels.

Example 5 Correlation analysis between DNA methylation level and age

Whole blood samples were collected from 241 healthy Chinese Han volunteers (128 males and 113 females) with an age range of 10-79 years, and blood stain samples were prepared and subjected to methylation analysis. Correlation analysis. The bloodstain age prediction model finally formed by the present invention includes 7 CpG sites spanning 3 genes, 3 known sites and 4 new CpG sites. The results showed that five of these CpG sites were derived from three genes (TRIM59, KLF14, and C1orf132), which were age-related in bloodstain analysis of Chinese subjects (Fig. 4).

Example 6 Comparison of Age Prediction Accuracy of Published Studies

A meta-analysis of age-prediction studies over the past several years showed that almost all MADs were >3 years old (Figure 5). Compared to previous studies, our model is the most efficient (MAD < 3 years with only 3-4 CpG sites) due to the use of RFR. In Figure 5, the solid dots represent published results, and the math in different models is represented by different shapes. The "ten" and "meter" symbols represent our age prediction results for women and men, respectively.

sequence listing

<110> Shanxi Medical University

<120> Methods for Age Prediction Using Pyrosequencing and Random Forest Regression Analysis

<160> 18

<170> SIPOSequenceListing 1.0

<210> 1

<211> 29

<212> DNA

<213> Artificial Sequence

<400> 1

tagtaaatat ataagtgggg gaagaaggg 29

<210> 2

<211> 27

<212> DNA

<213> Artificial Sequence

<400> 2

ttaataaaac caaattctaa aacattc 27

<210> 3

<211> 24

<212> DNA

<213> Artificial Sequence

<400> 3

caccttacca ccaaaccaaa attt 24

<210> 4

<211> 21

<212> DNA

<213> Artificial Sequence

<400> 4

aggggagtag ggtaagtgag g 21

<210> 5

<211> 30

<212> DNA

<213> Artificial Sequence

<400> 5

caaaaccatt tccccctaat atatacttca 30

<210> 6

<211> 20

<212> DNA

<213> Artificial Sequence

<400> 6

gggaggagat ttgtaggttt 20

<210> 7

<211> 21

<212> DNA

<213> Artificial Sequence

<400> 7

gggttttggg agtatagtag t 21

<210> 8

<211> 27

<212> DNA

<213> Artificial Sequence

<400> 8

acacctccta aaacttctcc aatctcc 27

<210> 9

<211> 21

<212> DNA

<213> Artificial Sequence

<400> 9

gttttgggag tatagtagtt a 21

<210> 10

<211> 28

<212> DNA

<213> Artificial Sequence

<400> 10

ggttttaggt taagttatgt ttaatagt 28

<210> 11

<211> 30

<212> DNA

<213> Artificial Sequence

<400> 11

actaaaaaat ttccctctat taccattacc 30

<210> 12

<211> 24

<212> DNA

<213> Artificial Sequence

<400> 12

atagttttag aaattatttt gttt 24

<210> 13

<211> 29

<212> DNA

<213> Artificial Sequence

<400> 13

tagtaaatat ataagtgggg gaagaaggg 29

<210> 14

<211> 28

<212> DNA

<213> Artificial Sequence

<400> 14

atttaataaa accaaattct aaaacatt 28

<210> 15

<211> 25

<212> DNA

<213> Artificial Sequence

<400> 15

ggggttaagt tattaagttt tgaag 25

<210> 16

<211> 21

<212> DNA

<213> Artificial Sequence

<400> 16

tataggtggt ttggggggaga g 21

<210> 17

<211> 27

<212> DNA

<213> Artificial Sequence

<400> 17

aaaaaacact accctccaca acataac 27

<210> 18

<211> 15

<212> DNA

<213> Artificial Sequence

<400> 18

ttgggggaga ggttg 15

Claims (4)

1. a method for age prediction, is characterized in that, comprises pyrosequencing and random forest regression analysis in described method, described random forest regression analysis model uses R package random Forest to build, and adopts positive selection. The best combination of loci was determined by the method; when the described random forest regression established the age prediction model, a total of 7 age-related DNA methylation loci were selected, of which 3 were male: TRIM59.pos7 , KLF14.pos2 , ELOVL2.pos7 ; 4 females are TRIM59.pos8 , KLF14.pos3 , Clorf132.pos2 and FHL2.pos6 ; The specific information of the DNA methylation sites is as follows: The chromosomal coordinate of the gene Clorf132 is chr1:207823675, and the CpG_ID is cg10501210; The chromosomal coordinate of the gene ELOVL2 is chr6:11044644, and the CpG_ID is cg16867657; The chromosome coordinate of the gene FHL2 is chr2:105399282, and the CpG_ID is cg06639320; The chromosomal coordinate of the gene KLF14 is chr7:130734355, and the CpG_ID is cg14361627; The chromosomal coordinate of the gene NPTX2 is chr7:98616518, and the CpG_ID is cg00548268; The chromosomal coordinate of the gene TRIM59 is chr3:160450199; The specific information of the DNA methylation site takes hg38 as the reference genome; The parameter settings in the random forest regression analysis model construction in the described method: the mtry parameter is the same as the number of CpG sites for each modeling, the minimum node size is 5, and the number of trees is set to 1000. 2. The method according to claim 1, wherein the volume of the PCR product used in the pyrosequencing is 12 μL. 3. the application of a group of methylation site combinations in age prediction, wherein the methylation site is made up of a male-related site and a female-related site; the male-related site is: TRIM59.pos7 , KLF14.pos2 , ELOVL2.pos7 ; female-related sites are: TRIM59.pos8 , KLF14.pos3 , Clorf132.pos2 , FHL2.pos6 ; The specific information of the DNA methylation sites is as follows: The chromosomal coordinate of the gene Clorf132 is chr1:207823675, and the CpG_ID is cg10501210; The chromosomal coordinate of the gene ELOVL2 is chr6:11044644, and the CpG_ID is cg16867657; The chromosome coordinate of the gene FHL2 is chr2:105399282, and the CpG_ID is cg06639320; The chromosomal coordinate of the gene KLF14 is chr7:130734355, and the CpG_ID is cg14361627; The chromosomal coordinate of the gene NPTX2 is chr7:98616518, and the CpG_ID is cg00548268; The chromosomal coordinate of the gene TRIM59 is chr3:160450199; The specific information of the DNA methylation site takes hg38 as the reference genome; The described age prediction is achieved by random forest regression analysis. 4. Use of the reagent for detecting the combination of methylation sites described in claim 3 in the preparation of a kit for predicting age.

Images