JOINT performs probability-based cell-type identification and DEG analysis for single-cell RNA sequencing (scRNA-seq) simultaneously without the need for imputation. It applies an EM algorithm on a generalized zero-inflated negative binomial mixture model. It supports arbitrary numbers of negative binomial components with and without zero inflation. JOINT performs soft-clustering by computing the probability of individual cells, i.e. each cell can belong to multiple cell types with different probabilities. The EM algorithm is implemented using the TensorFlow deep learning framework and can be run on both CPU and GPU.
There are several editorial errors in the online version of the JOINT paper. The corrected version can be found in JOINT_2021.pdf in this repo.
You can install JOINT after downloading this repo by running:
python setup.py install
joint solves the generalized zero-inflated negative binomial mixture model given counts.
joint(
data, # input array with row as gene and column as cell
K, # number of cell types
L = 2, # number of negative binomial components + 1
filter_genes = False, # whether do EM only on highly variable genes
n_top_genes=2000, # number of highly variable genes to keep if filter_genes=True
impute=True, # whether return imputation
n_inits=5, # number of initializations for the EM algorithm
n_init_iter=10, # number of runs of KMeans and Spectral clustering to initialize the EM algorithm
n_em_iter=100, # number of iterations to run the EM algorithm
n_inner_iter=50, # number of iterations to run the negative binomial inner loop inside the EM
tol=1e-5, # stop tolerance in EM algorithm
zero_inflated=True, # zero inflated or not
b_overwrites=[], # given beta in negative binomial components
normalize_data=True, # normalize data by library size
skip_spectral=True, # skip spectral clustering for initialization
) deg_unknown_labels DEG on two cell types without assuming true cell types are known.
deg_unknown_labels(
data, # input array with row as gene and column as cell
K, # number of cell types
k1, # cell type 1 for DEG
k2, # cell type 2 for DEG
em_res=None, # EM algorithm result. If None, it will run the EM algorithm first.
filter_genes = False, # whether do EM only on highly variable genes
n_top_genes=2000, # number of highly variable genes to keep if filter_genes=True
impute=True, # whether return imputation
n_inits=5, # number of initializations for the EM algorithm
n_init_iter=10, # number of runs of KMeans and Spectral clustering to initialize the EM
n_em_iter=100, # number of iterations to run the EM algorithm
n_inner_iter=50, # number of iterations to run the negative binomial inner loop inside the EM
tol=1e-5, # stop tolerance in EM algorithm
zero_inflated=True, # zero inflated or not
b_overwrites=[], # given beta in negative binomial components
normalize_data=True, # normalize data by library size
skip_spectral=True, # skip spectral clustering for initialization
) deg_known_labels DEG on two cell types assuming true cell types are known.
deg_known_labels(
data, # input array with row as gene and column as cell
K, # number of cell types
k1, # cell type 1 for DEG
k2, # cell type 2 for DEG
labels=None, # known cell types for each cell
em_res=None, # EM algorithm result. If None, it will run the EM algorithm first.
filter_genes = False, # whether do EM only on highly variable genes
n_top_genes=2000, # number of highly variable genes to keep if filter_genes=True
impute=True, # whether return imputation
n_inits=5, # number of initializations for the EM algorithm
n_init_iter=10, # number of runs of KMeans and Spectral clustering to initialize the EM algorithm
n_em_iter=100, # number of iterations to run the EM algorithm
n_inner_iter=50, # number of iterations to run the negative binomial inner loop inside the EM
tol=1e-5, # stop tolerance in EM algorithm
zero_inflated=True, # zero inflated or not
b_overwrites=[], # given beta in negative binomial components
normalize_data=True, # normalize data by library size
skip_spectral=True, # skip spectral clustering for initialization
) A notebook to show how to do EM based clustering can be found at examples/2celltype_clustering.ipynb
An example code to show how to use JOINT to do soft clustering and DEG can be found at examples/2celltype_sim.py
Cui, T., Wang, T. JOINT for large-scale single-cell RNA-sequencing analysis via soft-clustering and parallel computing. BMC Genomics 22, 47 (2021). https://doi.org/10.1186/s12864-020-07302-6
For technical issues particular to this repo, please report the issue on this GitHub repository.
New features, as well as bug fixes, by sending a pull request is always welcomed.
React
Typescript
Django
Laravel
Facebook
Microsoft
Google
Alibaba
D3
Tencent