labomics / DBP

DBP

DBP is a Python library for adaptive factor analysis of scRNA-seq data.

Usage examples

Generating training data for DBP

Rscript preprocess/pbmc.R 
Rscript preprocess/combine_subsets.R --task pbmc  && py preprocess/split_mat.py --task pbmc 

Training DBP

CUDA_VISIBLE_DEVICES=0 py run.py --exp alpha_13 --task pbmc

Comparison methods

task=pbmc
init=sp_00001899
K=50
Rscript comparison/liger.r --exp e0 --init_model $init --K $K --task $task

Quantitative evaluation

DBP

task=pbmc
init=sp_00001899
K=38
py eval/benchmark_batch_bio_break.py --task $task --init_model $init --K $K --method DBP

State-of-the-art methods

task=pbmc
init=sp_00001899
K=50
py eval/benchmark_batch_bio.py --task $task --init_model $init --K $K --method liger

File description

File or directory	Description
simulated datasets/	Script for generating simulated data sets
analysis/	Scripts for downstream analysis
comparison/	Scripts for algorithm comparison and qualitative evaluation
configs/	Dataset configuration and DBP model configuration
eval/	Scripts for quantitative evaluation
functions/	PyTorch functions for DBP
modules/	PyTorch models and dataloader for DBP
preprocess/	Scripts for data preprocessing
utils/	Commonly used functions
README.md	This file
run.py	Script for DBP training and inference