/DELAY

Depicting pseudotime-lagged causality for accurate gene-regulatory inference

Primary LanguagePythonGNU General Public License v3.0GPL-3.0

DELAY: DEpicting LAgged causalitY across single-cell trajectories for accurate gene-regulatory inference

DELAY

Quick Setup

  1. Follow these instructions to install the latest version of PyTorch with CUDA support: https://pytorch.org

    • Please note, DELAY currently requires CUDA-capable GPUs for training and prediction

  2. Confirm that two additional dependencies have been satisfied: pytorch-lightning and pandas

  3. Navigate to the location where you want to clone the repository and run:

git clone https://github.com/calebclayreagor/DELAY.git

Two Steps to Infer Gene-Regulatory Networks

1. Fine-tune DELAY on datasets with partially-known ground-truth interactions, e.g. from ChIP-seq experiments:

python RunDELAY.py [datadir] [outdir] -k [val_fold] [--atac] -p -ft
  • -k is the validation fold and --atac can optionally specify scATAC-seq input data (default is scRNA-seq)
  • Use TensorBoard to monitor training by runnning tensorboard --logdir RESULTS from the main directory
  • By default, DELAY will save the best model weights to a checkpoint file in RESULTS/outdir

2. Predict gene regulation across all TF-target gene pairs using the fine-tuned model:

python RunDELAY.py [datadir] [outdir] -m [RESULTS/outdir/BEST_WEIGHTS.ckpt] -p -g 1 -bs 1024
  • DELAY will save the predicted gene-regulation probabilities as a tfs x genes matrix in outdir named regPredictions.csv
  • By default, DELAY will load batches from existing directories, so make sure to delete created folders for all training, validation and prediction batches when finished

For additional help, run python RunDELAY.py --help

Required Input Files for Single-Cell Datasets

DELAY will expect unique sub-directories for each dataset in datadir containing the following files:

  1. NormalizedData.csv — A labeled genes x cells matrix of gene-expression or accessibility values

  2. PseudoTime.csv — A single-column table (cells x "PseudoTime") of inferred pseudotime values

  3. refNetwork.csv — A two-column table of ground-truth interactions between TFs ("Gene1") and target genes ("Gene2")

  4. TranscriptionFactors.csv (REQUIRED FOR INFERENCE) — A list of known transcription factors and co-factors in the dataset

  5. splitLabels.csv (REQUIRED FOR VALIDATION) — A single-column table (tfs x "Split") of training and validation folds for TFs in the refNetwork

For more help, see the example-data directory1

One Additional Example

Train a new VGG-6 model on datasets with fully-known ground-truth interactions:

python RunDELAY.py [datadir] [outdir] --train -k [val_fold] \
         --model_type vgg -cfg 32 32 M 64 64 M 128 128 M

Read the peer-reviewed paper: https://doi.org/10.1093/pnasnexus/pgad113

Footnotes

  1. Example data taken from Hayashi et al., Nature Communications (2018) ↩