A multi-tissue age prediction model based on DNA methylation analysis

Hussain Alsaleh, Nicola A. McCallum, Daniel L. Halligan, Penelope R. Haddrill

Research output: Contribution to journalArticle

1 Citation (Scopus)

Abstract

Age related tissue-specific DNA methylation markers have been identified in many studies, which can be used to estimate the chronological age of an unknown biological sample’s donor. However, if these markers have been used on the wrong type of tissue, they will give an inaccurate age estimation. This research has therefore examined HumanMethylation450 (HM450) BeadChip-based profiles retrieved from the NCBI repository, with the aim of identifying a set of universal DNA methylation markers across forensically relevant tissues. By using elastic net regression, it was possible to identify 10 age-related (AR) DNA methylation markers across 41 samples coming from five types of tissue (whole blood, saliva, semen, menstrual blood, and vaginal secretions). The average predictive accuracy of the constructed model based on training data is 3.8 years. In an independent dataset of 24 samples from four types of tissues (blood, saliva, menstrual blood, and vaginal secretions), the mean absolute deviation for the menstrual blood and vaginal fluid is 6.9 years, 5.6 years for buccal swabs, and 7.8 years for blood. The overall multi-tissue accuracy rate based on bootstrap analysis was 7.8 years (95% Confidence Interval 6–9.7 years). The identified multi-tissue age prediction model has the potential to assist forensic investigations without the requirement to identify the sample body fluid type.
LanguageEnglish
Pagese62-e64
Number of pages3
JournalForensic Science International: Genetics Supplement Series
Volume6
Early online date11 Sep 2017
DOIs
Publication statusPublished - 31 Dec 2017

Fingerprint

DNA Methylation
Genetic Markers
Saliva
Somatotypes
Cheek
Body Fluids
Semen
Confidence Intervals
Research

Keywords

  • DNA methylation
  • epigenetics
  • HM450
  • microarray
  • CpG sites
  • aging

Cite this

@article{fd3f8e6a7b44470b80d00159d08fb898,
title = "A multi-tissue age prediction model based on DNA methylation analysis",
abstract = "Age related tissue-specific DNA methylation markers have been identified in many studies, which can be used to estimate the chronological age of an unknown biological sample’s donor. However, if these markers have been used on the wrong type of tissue, they will give an inaccurate age estimation. This research has therefore examined HumanMethylation450 (HM450) BeadChip-based profiles retrieved from the NCBI repository, with the aim of identifying a set of universal DNA methylation markers across forensically relevant tissues. By using elastic net regression, it was possible to identify 10 age-related (AR) DNA methylation markers across 41 samples coming from five types of tissue (whole blood, saliva, semen, menstrual blood, and vaginal secretions). The average predictive accuracy of the constructed model based on training data is 3.8 years. In an independent dataset of 24 samples from four types of tissues (blood, saliva, menstrual blood, and vaginal secretions), the mean absolute deviation for the menstrual blood and vaginal fluid is 6.9 years, 5.6 years for buccal swabs, and 7.8 years for blood. The overall multi-tissue accuracy rate based on bootstrap analysis was 7.8 years (95{\%} Confidence Interval 6–9.7 years). The identified multi-tissue age prediction model has the potential to assist forensic investigations without the requirement to identify the sample body fluid type.",
keywords = "DNA methylation, epigenetics, HM450, microarray, CpG sites, aging",
author = "Hussain Alsaleh and McCallum, {Nicola A.} and Halligan, {Daniel L.} and Haddrill, {Penelope R.}",
year = "2017",
month = "12",
day = "31",
doi = "10.1016/j.fsigss.2017.09.056",
language = "English",
volume = "6",
pages = "e62--e64",
journal = "Forensic Science International: Genetics Supplement Series",
issn = "1875-1768",

}

A multi-tissue age prediction model based on DNA methylation analysis. / Alsaleh, Hussain; McCallum, Nicola A.; Halligan, Daniel L.; Haddrill, Penelope R.

In: Forensic Science International: Genetics Supplement Series, Vol. 6, 31.12.2017, p. e62-e64.

Research output: Contribution to journalArticle

TY - JOUR

T1 - A multi-tissue age prediction model based on DNA methylation analysis

AU - Alsaleh, Hussain

AU - McCallum, Nicola A.

AU - Halligan, Daniel L.

AU - Haddrill, Penelope R.

PY - 2017/12/31

Y1 - 2017/12/31

N2 - Age related tissue-specific DNA methylation markers have been identified in many studies, which can be used to estimate the chronological age of an unknown biological sample’s donor. However, if these markers have been used on the wrong type of tissue, they will give an inaccurate age estimation. This research has therefore examined HumanMethylation450 (HM450) BeadChip-based profiles retrieved from the NCBI repository, with the aim of identifying a set of universal DNA methylation markers across forensically relevant tissues. By using elastic net regression, it was possible to identify 10 age-related (AR) DNA methylation markers across 41 samples coming from five types of tissue (whole blood, saliva, semen, menstrual blood, and vaginal secretions). The average predictive accuracy of the constructed model based on training data is 3.8 years. In an independent dataset of 24 samples from four types of tissues (blood, saliva, menstrual blood, and vaginal secretions), the mean absolute deviation for the menstrual blood and vaginal fluid is 6.9 years, 5.6 years for buccal swabs, and 7.8 years for blood. The overall multi-tissue accuracy rate based on bootstrap analysis was 7.8 years (95% Confidence Interval 6–9.7 years). The identified multi-tissue age prediction model has the potential to assist forensic investigations without the requirement to identify the sample body fluid type.

AB - Age related tissue-specific DNA methylation markers have been identified in many studies, which can be used to estimate the chronological age of an unknown biological sample’s donor. However, if these markers have been used on the wrong type of tissue, they will give an inaccurate age estimation. This research has therefore examined HumanMethylation450 (HM450) BeadChip-based profiles retrieved from the NCBI repository, with the aim of identifying a set of universal DNA methylation markers across forensically relevant tissues. By using elastic net regression, it was possible to identify 10 age-related (AR) DNA methylation markers across 41 samples coming from five types of tissue (whole blood, saliva, semen, menstrual blood, and vaginal secretions). The average predictive accuracy of the constructed model based on training data is 3.8 years. In an independent dataset of 24 samples from four types of tissues (blood, saliva, menstrual blood, and vaginal secretions), the mean absolute deviation for the menstrual blood and vaginal fluid is 6.9 years, 5.6 years for buccal swabs, and 7.8 years for blood. The overall multi-tissue accuracy rate based on bootstrap analysis was 7.8 years (95% Confidence Interval 6–9.7 years). The identified multi-tissue age prediction model has the potential to assist forensic investigations without the requirement to identify the sample body fluid type.

KW - DNA methylation

KW - epigenetics

KW - HM450

KW - microarray

KW - CpG sites

KW - aging

UR - http://www.sciencedirect.com/science/journal/18751768

U2 - 10.1016/j.fsigss.2017.09.056

DO - 10.1016/j.fsigss.2017.09.056

M3 - Article

VL - 6

SP - e62-e64

JO - Forensic Science International: Genetics Supplement Series

T2 - Forensic Science International: Genetics Supplement Series

JF - Forensic Science International: Genetics Supplement Series

SN - 1875-1768

ER -