SMILE: Mutual Information Learning for Integration of Single Cell Omics Data
Xu et al. "SMILE: mutual information learning for integration of single-cell omics data". Bioinformatics
- numpy
- scipy
- pandas
- scikit-learn
- scanpy
- pytorch
To identify shared signatures across modalities via SMILE, please see tutorial
|----SMILE_identify_shared_signatures_across_modalities.ipynb
##rna_X: RNA-seq data; dna_X: ATAC-seq data, rna_X and dna_X are paired data
##rna_X_unpaired: RNA-seq data; dna_X_unpaired: ATAC-seq data, rna_X_unpaired and dna_X_unpaired are unpaired data, and we wish to integrate unpaired data
##Both rna_X and dna_X are matrices in which each row represents one cell while each column stands for a feature
##each row in rna_X and dna_X is paired for training purpose
##Within modality, for example rna_X and rna_X_unpaired, data share the same feature space
## Proecessed data could be found at https://doi.org/10.5281/zenodo.7776066
from SMILE import littleSMILE
from SMILE import ReferenceSMILE_trainer
integrater = littleSMILE(input_dim_a=rna_X.shape[1],input_dim_b=dna_X.shape[1],clf_out=20)
ReferenceSMILE_trainer(rna_X,dna_X,rna_X_unpaired,dna_X_unpaired, integrater, train_epoch=1000)
Integrate RNA-seq/ATAC-seq from Granja et al 2019 with 10X multiome PBMC data as reference
|----SMILE_data_integration_withReference.ipynb
##rna_X: RNA-seq data; dna_X: ATAC-seq data
##Both rna_X and dna_X are matrices in which each row represents one cell while each column stands for a feature
##each row in rna_X and dna_X is paired for training purpose
from SMILE import SMILE
from SMILE import PairedSMILE_trainer
net = littleSMILE(input_dim_a=rna_X.shape[1],input_dim_b=dna_X.shape[1],clf_out=30)
ReferenceSMILE_trainer(X_rna_paired,X_dna_paired,X_rna_unpaired,X_dna_unpaired,net,batch_size=1024, f_temp = 0.2)
For use of SMILE in multi-source single-cell transcriptome data
|----SMILE_MouserCortex_RNAseq_example.ipynb
For use of SMILE in integration of multimodal single-cell data
|----SMILE_Celllines_RNA-ATAC-integration_example.ipynb
|----screen_factors_for_coembedding.py
Processed data can be found at https://figshare.com/articles/dataset/Mouse_skin_data_by_SHARE-seq/16620367
React
Typescript
Django
Laravel
Facebook
Microsoft
Google
Alibaba
D3
Tencent