CellDART is a tool to estimate cell fraction of spatial transcriptomic spots using domain adaptation of deep neural networks.

• Cite as: Bae S, Na KJ, Koh J, Lee DS, Choi H, Kim YT. CellDART: cell type inference by domain adaptation of single-cell and spatial transcriptomic data. Nucleic Acids Res. 2022;50(10):e57.
  

Number of total marker genes = 200 ~ 400 (or number of markers per cluster: 10 ~ 20)
Number of pseudospots = 5 to 10 times the number of real spots (20,000~40,000 per Visium slide)
Number of sampled cells in a pseudospot (virtual mixture of single-cell data) = 8
Iteration number = 3,000
Mini-batch size = 512
Loss weights between source and domain classifier (alpha) = 0.6
Learning rate = 0.001 * alpha_lr = 0.005

python: CellDART_example_mousebrain_markers.ipynb
-> example file that shows the whole process step by step
-> the pred_cellf_celldart function is a compressed version of all the steps shown in the notebook file
-> see the below example how to use the function
R wrap: Please refer to the '/vignettes/introduction.Rmd' file

  conda create -n CellDART python=3.8  
  conda activate CellDART  
  pip install git+https://github.com/mexchy1000/CellDART.git  
  python -m ipykernel install --user --name CellDART --display-name CellDART  

python 3.8
tensorflow 2.9.1
numpy 1.20.3
pandas 1.4.3
scanpy 1.9.1
jupyter 1.0.0

from CellDART.pred_cellf_celldart import pred_cellf_celldart  
adata_sp = pred_cellf_celldart(adata_sp=adata_sp, adata_sc=adata_sc, count_from_raw = False,  
        　　　　　　　　　　　   gpu=True, celltype='celltype', num_markers=20,  
        　　　　　　　　　　　　 nmix=8, npseudo=20000, alpha=0.6, alpha_lr=5, batch_size=512,  
        　　　　　　　　　　　　 emb_dim=64, n_iterations=3000, init_train_epoch=10,  
        　　　　　　　　　　　　 outdir='./CellDART_output', return_anndata=True)

(1) adata_sp: spatial data (AnnData object) with raw count matrix to be used in predicting cell fraction (default: None)
(2) adata_sc: single-cell data (AnnData object) with raw count matrix to be used in making pseudospots (default: None)
(3) count_from_raw: whether to extract count matrix frow .raw of AnnData (default: False)
-> non-normalized raw count matrix should be contained in the AnnData .raw file
-> if False, then utilize the count matrices saved in adata_sp and adata_sc directly
(4) gpu: check whether to use gpu (True) or not (False) (default = True)
(5) celltype: column name for single-cell annotation data in .obs (default: 'celltype')
(6) num_markers: number of selected marker genes in each celltype (default = 20)
(7) nmix: sampling number of cells in pseudospot (default = 10)
(8) npseudo: a total number of pseudospots (default = 20,000)
(9) alpha: loss weights of the domain classifier to the source classifier (default = 0.6)
(10) alpha_lr: learning rate for the domain classifier (alpha_lr*0.001, default = 5)
(11) batch_size: minibatch size for pseudospots and spatial data during the training (default = 512)
(12) n_iterations: iteration number for the adversarial learning (default = 3,000)
(13) init_train_epoch: iteration number for the pre-training process (default = 10)
(14) outdir: the directory to save output files (models and results)
(15) return_anndata: whether to return spatial AnnData file with predicted cell fraction in .obs (default: False)

devtools::install_github("mexchy1000/CellDART", build_vignettes = T, force = T)  
library(CellDART)  
help(pred_cellf_celldart)  # Explanation for the parameters and short examples  
browseVignettes("CellDART")  # Browse for the vignettes (/vignettes/introduction.Rmd)

# Using conda environment (environment will be automatically installed in Linux distributions)
# If using Windows, then install conda environment first and then run the function below with python.install = F
sp_data_cellf <- pred_cellf_celldart(sp_data, sc_data, outdir = '.',
                                     sp_subset=F, spot.cluster.name='seurat_clusters',
                                     spot.cluster.of.interest=NULL,
                                     env.select='conda',python.install=T,
                                     python_path=NULL, env.name='CellDART',
                                     gpu=TRUE, metadata_celltype='celltype',
                                     num_markers=20, seed_num=0,
                                     nmix=8, npseudo=20000, alpha=0.6,alpha_lr=5,
                                     emb_dim=64,batch_size=512,n_iterations=3000, init_train_epoch=10)

# Using virtual environment (environment will be automatically installed in Linux distributions)
# Not recommended for Windows
sp_data_cellf <- pred_cellf_celldart(sp_data, sc_data, outdir = '.',
                                     sp_subset=F, spot.cluster.name='seurat_clusters',
                                     spot.cluster.of.interest=NULL,
                                     env.select='virtual',python.install=T,
                                     python_path=NULL, env.name='CellDART',
                                     gpu=TRUE, metadata_celltype='celltype',
                                     num_markers=20, seed_num=0,
                                     nmix=8, npseudo=20000, alpha=0.6,alpha_lr=5,
                                     emb_dim=64,batch_size=512,n_iterations=3000, init_train_epoch=10)

(1) outdir: the directory to save output files (models and results) (default = '.')
(2) sp_subset: whether to subset spatial data and calculate for specific spot cluster (default = FALSE)
(3) spot.cluster.name: group name of the cluster used for subsetting spatial data (default = 'seurat_clusters')
(4) spot.cluster.of.interest: name of each spot clusters to be used (default = NULL)
(5) env.select: select between using reticulate virtual environment or conda environment (default = 'conda')
-> either of the selection will search the already installed environment
-> if environment is not found, then it will automatically install the new environment
(6) python.install: whether to automatically install python version 3.8.13 (default = F)
-> For Windows, set python.install = F
(7) python_path: path for the python 3.8.13 (default = NULL)
(8) env.name: name of the virtual or conda environment to use for the analysis (default = 'CellDART')
(9) metadata_celltype: column name for single-cell annotation data in metadata (default = 'celltype')

Seurat 4.0.5  
dplyr 1.0.7  
sceasy 0.0.6  
reticulate 1.22  

Virtual environment (env.select="virtual") or conda environment (env.select="conda") will be automatically installed while running function 'pred_cellf_celldart'
Detailed explanation is in '/R/Read_R_wrap.md' file.

Install conda environment first and then run the function with env.select='conda' and python.install=F

Shiny application for preprocessing and CellDART analysis. (inside 'shiny')
Refer to the STquantool application: STquantool