tAge

R package for transcriptomic biological age prediction from gene expression data.

Installation

devtools::install_github("Gladyshev-Lab/tAge")

Python dependency

tAge uses Python for model prediction. Set up a virtual environment with required packages:

python -m venv .venv
.venv/bin/pip install joblib pandas scikit-learn

Then in R, before running predictions:

Sys.setenv(RETICULATE_PYTHON = ".venv/bin/python")

Quick start

Bulk RNA-seq

library(tAge)

# Load data
exprs_data <- read.csv("expression_matrix.csv", row.names = 1)
metadata <- read.csv("metadata.csv", row.names = 1)
eset <- make_ExpressionSet(exprs_data, metadata)

# Preprocess
tAge_eset <- tAge_preprocessing(
  eset,
  species = "mouse",
  gene_mapping_type = "Gene.Symbol",
  control_group_column = "treatment",
  control_group_label = "control"
)

# Predict
model_paths <- list(
  scaled_diff = "path/to/EN_scaleddiff.pkl",
  yugene_diff = "path/to/EN_yugenediff.pkl"
)
Sys.setenv(RETICULATE_PYTHON = ".venv/bin/python")

results <- predict_tAge(tAge_eset, model_paths, species = "mouse", mode = "EN")

Single-cell (pseudobulk)

library(Seurat)

# Pseudobulk aggregation by sample and tissue
eset <- aggregate_on_obs_columns(
  seurat_obj,
  obs_column_names = c("sample_id", "tissue"),
  coverage_threshold = 1e7
)

# Remove outliers
eset_clean <- remove_outliers(eset, split_by = "tissue")

# Run tAge per tissue
results <- tAge_by_group(
  eset_clean,
  split_by = "tissue",
  model_paths = model_paths,
  species = "mouse",
  mode = "EN",
  control_group_column = "age",
  control_group_label = "young"
)

Supported species

Mouse, human, rat, monkey. Non-mouse species are mapped to mouse orthologs internally.

Citation

Transcriptomic Hallmarks of Mortality Reveal Universal and Specific Mechanisms of Aging, Chronic Disease, and Rejuvenation (Tyshkovskiy et al. 2024)

License

MIT