Designing epigenetic clocks for wildlife research
Data files
Apr 22, 2025 version files 599.78 MB
-
full_sibs.txt
508 B
-
iscans.zip
599.31 MB
-
low_qc_positions.csv
517 B
-
PB_aligned_cpgs.txt
437.93 KB
-
README.md
4.27 KB
-
sample_sheets.zip
27.25 KB
Abstract
The potential applications of epigenetic clocks are expanding in wildlife conservation and management. The pace at which they are being adopted highlights the need for field-specific design best practices. Epigenetic clocks were originally developed for human studies, presenting challenges for their adoption in wildlife research. Most notably, the estimated ages of sampled wildlife can be unreliable, and sampling restrictions limit the number and variety of available samples, which can reduce the accuracy of epigenetic clocks for wildlife. In this article, we present a detailed workflow for designing, validating, and applying wildlife epigenetic clocks in a way that accounts for sampling constraints. We provide recommendations for two main applications of wildlife epigenetic clocks: estimating unknown ages and assessing cumulative biological aging. Our simulations and analyses, applied to an extensive polar bear dataset from across the Canadian Arctic, demonstrate that accurate epigenetic clocks for wildlife can be constructed and validated with limited samples, accommodating projects with small budgets and sampling constraints. With our workflow and examples, we hope to make epigenetic clocks more accessible and widespread in wildlife conservation and management.
Dataset DOI: 10.5061/dryad.rxwdbrvmw
Description of the data and file structure
The data include sample sheets, raw idat files, and quality-control information that can be used to process the DNA methylation data into a suitable format for developing epigenetic clocks.
Files and variables
File: low_qc_positions.csv
Description: Low-quality detection p-values (high values indicate quality issues) for screening samples.
Variables
- position: chip position
- chip.ID.loc: chip IDs corresponding to the name of the position (corresponds to chip.ID_stripe columns from the sample sheets)
- detection_p: detection p-value; > 0.05 is high
- batch_no: Identifier ranging from 1-8
File: full_sibs.txt
Description: Individual sample IDs (corresponding to sampleId in sample sheets) with siblings in the dataset. These individuals are normally excluded from clock design because of potential bias due to genetic relatedness.
File: PB_aligned_cpgs.txt
Description: Names of cpg sites (corresponding to sites from Horvath Mammal Array in idat files) that align with the polar bear genome.
File: sample_sheets.zip
Description: Contains files with biological information about the samples and technical information about sample locations on the arrays.
- PB_arrayXX_sample_sheetXX.rds: Data frame with information about the array location corresponding to each sample in batchXX_samples.rds.
- Sample_Name: Unique sample name (corresponds to sampleId in batchXX_samples.rds)
- Sample_Well: Location on 96-well plate
- Sample_Plate: Identifies the plate for batches with multiple plates
- chip.No: Unique identifier for the chip (four per array)
- chip.ID: Numeric identifier for the chip (four per array)
- stripe: Row and column position of the sample on the chip (RXX = rows 1-6, CXX = columns 1-2)
- row: Row identifier for 96-well plate (8 rows, a-h)
- column: Column identifier for 96-well plate (12 rows, numbered)
- batchXX_samples.rds: Data frame with biological sample information corresponding to each sample in the sample sheets. Note: NA indicates missing values. Missing values in the ID column indicate bears that were not assigned an individual bear ID at the time of sampling. This happened when bears were sampled as part of necropsies, meaning they did not require an identifier for future sampling. For some of these necropsied bears, ages were also unknown, resulting in missing values in the age and Born columns.
- sampleId: Unique sample name
- id: ID for individual bear (might be multiple sampleIds/bear)
- Population: Population code (SB = Southern Beaufort, NB = Northern Beaufort, GB = Gulf of Boothia, LS = Lancaster Sound, MC = Mc'Clintock Channel, FB = Foxe Basin, WH = Western Hudson Bay, SH = Southern Hudson Bay, DS = Davis Strait, BB = Baffin Bay)
- Spec: Tissue type (blood, skin, muscle)
- YMD: Year-month-day of sample
- sex: Sex of individual (F = female, M = male)
- age: Age (in years) of the individual at the time of the sample
- Born: Birth year of the individual
File: iscans.zip
Description: Contains idat files arranged in folders labeled with batches corresponding to numbered sample sheet files, each containing folders corresponding to chip.ID columns in the sample sheets, containing red and green channel .idat files named with chip position (stripe column in sample sheet files).
Code/software
All code was run in R v4.3.1 and runs on macOS Monterey v12.7.3
CRAN Package versions:
- tidyverse v2.0.0
- cowplot v1.1.3
- glmnet v4.1-8
- limma v1.1.3
- BiocManager v1.30.22
Developmental Package versions:
- packages/HorvathMammalMethylChip40anno.test.unknown_0.2.2.tar.gz
- packages/HorvathMammalMethylChip40manifest_0.2.2.tar.gz
Note that both package versions were downloaded and are available from https://github.com/shorvath/MammalianMethylationConsortium
Access information
Other publicly accessible locations of the data:
We compiled an extensive DNA methylation dataset from polar bears across the Canadian Arctic to assess whether sampling biases, data pre-processing, and validation influence wildlife clock performance. Our dataset includes DNA from 10 distinct subpopulations, each with different proportions of blood, skin, and muscle tissue, from male and female bears, and representing ages across the typical lifespan of a wild polar bear from age 0 to 30.
Scripts for processing the data are available from the GitHub repository https://github.com/ljnewediuk/how_to_clocks.git.
