This repository contains the code used to analyze the genetics of aging and lifespan in a study on a large cohort of UM-HET3 mice. The scripts map loci that influence mortality, revealing how their effects change with age and sex. The code also investigates the genetic basis for the relationship between body weight and lifespan, identifying distinct sets of genes that control this correlation at different life stages.

The repository provides the tools for reproducing the findings presented in the corresponding paper. It includes scripts for the actuarial analysis of survival data, the mapping of Vita loci that influence lifespan, and the identification of Soma loci that control the relationship between body weight and longevity. The code also supports the analysis of epistatic interactions between these genetic loci, highlighting the complex genetic architecture of aging.

Most scripts in this repository start by loading in the um-het3-rqtl.csvr, which is the data coded for R/qtl-v1 as explained in the read.cross help file. The um-het3-rqtl.csvr file can be loaded in with the R/qtl-v1 read.cross function, with parameters set to genotypes=NULL and na.strings=c("-", "NA")

The folder PreCross/ contains all the code used to call SNPs on the monsterplex BAM files, and the conversion & harmonization of Monsterplex and Sequenom data into the cross object. It also contains the code to convert observed SNPs to the Founder strain haplotypes. This code is used to produce the um-het3-rqtl.csvr from Monsterplex .BAM files and the Sequenom data, with phenotypes and covariates coming from genenetwork.org.

The folder ProgessiveMapping/ contains all the code used to perform actuarial QTL scans on lifespan for 4-way, paternal and maternal maps used to discover Vita loci. This folder also contains the adjustXprobs.R code, that is needed for proper X-chromosome mapping, and furthermore contains bodyweight QTL mapping as well as simpleM and Cauchi Combination tests for multiple testing correction. The Lasso based non-parametric Quantile Regression 'engine' code is also found here.

The folder Soma/ contains all the code used to perform the Correlated Trait Locus (CTL) mapping analysis between bodyweight (at 5 timepoints) and progressive lifespan mapping. Furthermore, code for visualizing Soma loci is also located in this folder.

The folder Interactions/ contains the interaction models used to investigate / map all interactions between G x G, G x Sex, G x Site, and G x Drug. As well as code for creating the big interaction table figure inside the paper.

The folder Figures/ contains code that produces the figures / images inside the paper. Not all figure code is found here some code lives in their respective folders (e.g. visualizations of predictions are found in Predictions/, while some GxG images are found in Interactions/.

The folder Inversions/ contains the inversion finder code, which finds inversion by looking at recombination frequency using an 8-marker sliding window.

The folder CandidateGenes/ contains all the code used to investigate and prioritize genetic features and possible candidate genes, code for biomaRt, VEP prediction, pathway over-representation analysis (ORA), and cross referencing genes in Vita / Soma regions with homologs in GenAge known to influence lifespan is found in this folder.

The folder SingleLocus/ contains all the code used to investigate single loci in the UM-HET3 genome (e.g. The Vita9B locus previously found by R. Miller), as well as code for mapping the control animals used in the Science paper by Sleiman et al. to compare/align their original work on a subset of animals with our mapping approach. Furthermore, there is code to investigate distal chromosome X and proximal chromosome 1, at which we observed some distortion in recombination frequency.

The folder Various/ contains all the code for computing main effect and epistatic heritability, as well code for the adaptLoess approach (and figures) and other various scripts that do not neatly fit into any other folder.

The folder Predictions/ contains all the code used to predict lifespan either by using random forest on covariates, as well as our two novel approaches ways of performing polygenic risk prediction (not in the paper, but in the biorxiv version).

The folder Old/ contains the old & deprecated code not used anymore, or which was used to do one off analysis.

The code in this ropsitory is made possible by, and has dependencies on, the following excellent software:

• The R Project for Statistical Computing
• Ensembl Variant Effect Predictor (Ensembl VEP)
• biomaRt
• R/qtl
• Bcftools

As well as many many different R-packages used in the analysis of the data.

The code for progressive QTL mapping in this repository follows specific conventions to ensure consistency and clarity. The analysis relies on a genetic map created from Monsterplex Capture DNA-Sequencing and Sequenom MassARRAY data. The codebase is designed to handle the complex genetic structure of the UM-HET3 population, a four-way cross derived from four distinct parental strains.

The data is formatted for R/qtl-v1, and as such genotypes are coded as follows:

• AA for the BALB/cByJ strain
• BB for the C57BL/6J strain
• CC for the C3H/HeJ strain
• DD for the DBA/2J strain

The code also accounts for phase-known alleles where the counterpart is uncertain or not imputed. These are represented with a question mark. For example, a genotype with an A allele from the BALB/cByJ parent and an unknown allele from the other is coded as A?. Similarly, an unknown allele combined with a known C allele is represented as ?C. This convention allows the analysis to proceed even with incomplete genotyping data, maintaining the integrity of the genetic map.

Want to contribute? Great! Contribute to this repo by starring ⭐ or forking 🍴, and feel free to start an issue first to discuss idea's before sending a pull request. You're also welcome to post comments on commits.

Or be a maintainer, and adopt (the documentation of) a function.

Written by Danny Arends and is released under the GNU GENERAL PUBLIC LICENSE Version 3 (GPLv3). See LICENSE.txt.

Arends, D., Ashbrook, D. G., Roy, S., Lu, L., Sloan, Z., Centeno, A. G., Lamour, K. H., de Magalhães, J. P., Prins, P., Broman, K. W., Sen, S., Mitchell, S. J., MacArthur, M. R., Akin, Ö. A., Auwerx, J., Bajwa, A., Diaz, V., Harrison, D. E., Strong, R., Nelson, J. F., Mozhui, K., Williams, E. G., Miller, R. A., & Williams, R. W. (2025). Genetic Modulation of Lifespan: Dynamic Effects, Sex Differences, and Body Weight Trade-offs bioRxiv https://doi.org/10.1101/2025.04.27.649857