Learn R Programming

celda: CEllular Latent Dirichlet Allocation

"celda" stands for "CEllular Latent Dirichlet Allocation", which is a suite of Bayesian hierarchical models and supporting functions to perform gene and cell clustering for count data generated by single cell RNA-seq platforms. This algorithm is an extension of the Latent Dirichlet Allocation (LDA) topic modeling framework that has been popular in text mining applications. Celda has advantages over other clustering frameworks:

Celda can simultaneously cluster genes into transcriptional states and cells into subpopulations
Celda uses count-based Dirichlet-multinomial distributions so no additional normalization is required for 3' DGE single cell RNA-seq
These types of models have shown good performance with sparse data.

Installation Instructions

To install the most recent beta release of celda via devtools:

library(devtools)
install_github("compbiomed/celda@v0.3")

The most up-to-date (but potentially less stable) version of celda can similarly be installed with:

install_github("compbiomed/celda")

Examples and vignettes

Vignettes are available in the package.

An analysis example using celda with RNASeq via vignette('celda-analysis')

New Features and announcements

The v0.2 release of celda represents a useable implementation of the various celda clustering models. Please submit any usability issues or bugs to the issue tracker at https://github.com/compbiomed/celda

You can discuss celda, or ask the developers usage questions, in our Google Group.

Copy Link

Version

Version

0.0.0.9000

License

MIT

Maintainer

Joshua Campbell

Last Published

August 8th, 2018

Functions in celda (0.0.0.9000)

cluster_probability.celda_C

Calculates the conditional probability of each cell belong to each cluster given all other cluster assignments

getK for celda Gene clustering model

completeClusterHistory

Get the complete history of gene / cell / gene & cell cluster assignments generated during a celda run, dependent on the model provided.

finalClusterAssignment.celda_CG

finalClusterAssignment for the celda Cell and Gene clustering model

distinct_colors

Generate a distinct palette for coloring different clusters

Create a scatterplot given two dimensions from a data dimensionality reduction tool (e.g tSNE)

celda_heatmap.celda_CG

celda_heatmap for celda Cell and Gene clustering model

Create a scatterplot for each row of a normalized gene expression matrix where x and y axis are from a data dimensionality reduction tool.

Get the L value used for each chain in a celda run.

celda_heatmap.celda_G

celda_heatmap for celda Gene clustering model

getK for the celda Cell and Gene clustering model

completeLogLikelihood

Get the complete log likelihood for a given celda model.

getK for celda Cell clustering function

normalizeCounts

Normalize a counts matrix by a scalar factor

visualize_model_performance

Visualize various performance metrics as a function of K / L to aid parameter choice.

plot_dr_cluster

Create a scatterplot based on celda cluster labels.

visualize_model_performance.celda_C

visualize_model_performance for celda Cell clustering function

A function to draw clustered heatmaps.

factorizeMatrix.celda_C

Generate factorized matrices showing each feature's influence on the celda_C model clustering

getL for celda Cell clustering function

getL for the celda Cell and Gene clustering model

finalLogLikelihood

Get the log likelihood from the final iteration of Gibbs sampling for a given celda model.

render_celda_heatmap

Render a heatmap based on a matrix of counts where rows are genes and columns are cells.

finalClusterAssignment.celda_G

finalClusterAssignment for celda Gene clustering model

Get the K value used for each chain in a celda run.

render_interactive_kl_plot

Render an interactive figure demonstrating the clustering performance of different K/L parameters

celda_heatmap.celda_C

celda_heatmap for celda Cell clustering function

factorizeMatrix.celda_CG

Generate factorized matrices showing each feature's influence on the celda_CG model clustering

Simulate count data from the celda generative models.

factorizeMatrix.celda_G

Generate factorized matrices showing each feature's influence on the celda_G model clustering

getL for celda Gene clustering model

render_iteration_likelihood_plot

Visualize log likelihood per iteration of Gibbs sampling

Create a scatterplot based off of a matrix containing the celda state probabilities per cell.

Get run parameters for a celda run.

Select the best model amongst a list of models generated in a celda run.

Get the random seed for a given celda model.

recode_cluster_labels

Re-code cluster label by provided mapping scheme

simulateCells.celda_G

Simulate cells from the gene clustering generative model

simulateCells.celda_C

Simulate cells from the cell clustering generative model

Identify features with the highest influence on clustering

visualize_model_performance.celda_CG

Visualize the performance of celda_CG models grouped by L and K

simulateCells.celda_CG

Simulate cells from the cell/gene clustering generative model

visualize_model_performance.celda_G

visualize_model_performance for the celda Gene function

calculate_loglik_from_variables.celda_CG

Calculate log lileklihood for the celda Cell and Gene clustering model, given a set of cell / gene cluster assignments

celda Cell and Gene Clustering Model

available_models

available models

celda Cell Clustering Model

calculate_marginal_likelihood

Calculate the marginal likelihood from a single celda model

calculate_perplexity

Calculate the perplexity from a single celda chain

calculate_loglik_from_variables.celda_G

Calculate the celda_G log likelihood for user-provided cluster assignments

clusterProbabilities

Get the probability of the cluster assignments generated during a celda run.

calculate_loglik_from_variables

Calculate a log-likelihood for a user-provided cluster assignment and count matrix, per the desired celda model.

calculate_loglik_from_variables.celda_C

Calculate the celda_C log likelihood for user-provided cluster assignments

Run the celda Bayesian hierarchical model on a matrix of counts.

celda Gene Clustering Model

chooseBestChain

Determine the best chain among a set of celda_* objects with otherwise uniform K/L choices.

clusterProbabilities.celda_C

clusterProbabilities for celda Cell clustering function

Render a stylable heatmap of count data based on celda clustering results.

compare_count_matrix

Check whether a count matrix was the one used in a given celda run

completeClusterHistory.celda_G

completeClusterHistory for celda Gene clustering model

completeClusterHistory.celda_C

completeClusterHistory for celda Cell clustering funciton

finalClusterAssignment.celda_C

finalClusterAssignment for celda Cell clustering funciton

completeClusterHistory.celda_CG

completeClusterHistory for the celda Cell and Gene clustering model

clusterProbabilities.celda_CG

clusterProbabilities for the celda Cell and Gene clustering model

clusterProbabilities.celda_G

clusterProbabilities for celda Gene clustering model

finalClusterAssignment

Get the final gene / cell / gene & cell cluster assignments generated during a celda run, dependent on the model provided.

Render a stylable heatmap of count data based on celda clustering results.

calculateLoglikFromVariables.celda_C

Calculate the celda_C log likelihood for user-provided cluster assignments

calculateLoglikFromVariables.celda_CG

Calculate log lileklihood for the celda Cell and Gene clustering model, given a set of cell / gene cluster assignments

celdaHeatmap.celda_CG

celdaHeatmap for celda Cell and Gene clustering model

calculateLoglikFromVariables.celda_G

Calculate the celda_G log likelihood for user-provided cluster assignments

clusterProbability.celda_C

Calculates the conditional probability of each cell belong to each cluster given all other cluster assignments

clusterProbability.celda_CG

Calculates the conditional probability of each cell belong to each cluster given all other cluster assignments

calculatePerplexity

Calculate the perplexity from a single celda chain

clusterProbability.celda_G

Calculates the conditional probability of each cell belong to each cluster given all other cluster assignments

celdaHeatmap.celda_G

celdaHeatmap for celda Gene clustering model

compareCountMatrix

Check whether a count matrix was the one used in a given celda run

celdaHeatmap.celda_C

celdaHeatmap for celda Cell clustering function

calculateLoglikFromVariables

Calculate a log-likelihood for a user-provided cluster assignment and count matrix, per the desired celda model.

Plots the log likehood over iteration for each chain

Create a scatterplot given two dimensions from a data dimensionality reduction tool (e.g tSNE)

factorizeMatrix

Generate factorized matrices showing each feature's influence on cell / gene clustering

renderCeldaHeatmap

Render a heatmap based on a matrix of counts where rows are genes and columns are cells.

Re-code cell cluster labels by provided mapping scheme

Create a scatterplot for each row of a normalized gene expression matrix where x and y axis are from a data dimensionality reduction tool.

Select a set of models from a celda_list objects based off of rows in its run.params attribute.

visualizeModelPerformance

Visualize various performance metrics as a function of K / L to aid parameter choice.

Create a scatterplot based off of a matrix containing the celda state probabilities per cell.

visualizeModelPerformance.celda_C

visualizeModelPerformance for celda Cell clustering function

Create a scatterplot based on celda cluster labels.

Re-code gene cluster labels by provided mapping scheme

visualizeModelPerformance.celda_CG

Visualize the performance of celda_CG models grouped by L and K

renderInteractiveKLPlot

Render an interactive figure demonstrating the clustering performance of different K/L parameters

renderIterationLikelihoodPlot

Visualize log likelihood per iteration of Gibbs sampling

visualizeModelPerformance.celda_G

visualizeModelPerformance for the celda Gene function

cCG.decomposeCounts

Takes raw counts matrix and converts it to a series of matrices needed for log likelihood calculation

cC.decomposeCounts

Takes raw counts matrix and converts it to a series of matrices needed for log likelihood calculation

cG.decomposeCounts

Takes raw counts matrix and converts it to a series of matrices needed for log likelihood calculation

absoluteProbabilityHeatmap

Renders a heatmap based on a population matrix from the factorized counts matrix.

Create a bar chart based on Gini coefficients of transcriptional states

Runs tSNE via Rtsne based on the CELDA model and specified cell states.

eigenMatMultInt

Fast matrix multiplication for double x int

clusterProbability

Get the probability of the cluster assignments generated during a celda run.

calculatePerplexityWithResampling

Calculate and visualize perplexity of all models in a celda_list, with resampling

calc.perplexity

calc.perplexity

calc.perplexity2

calc.perplexity2

gettingClusters

Assists in selection of K/L parameter for downstream analysis.

lookupGeneModule

Obtain the gene module of a gene of interest

diffExpBetweenCellStates

Gene expression markers for cell clusters using MAST

relativeProbabilityHeatmap

Renders a heatmap based on a population matrix from the factorized counts matrix. The relative probability of each transcriptional state in each cell subpopulation is visualized.

Transcriptional state heatmap

Select the best model from a celda_list, as determined by final log-likelihood.

validateRunParams

Check whether the given celda_list's run.params attribute ordering matches it's res.list models

visualizePerplexity

Visualize perplexity as a function of cell/gene clustering

cluster_probability.celda_CG

Calculates the conditional probability of each cell belong to each cluster given all other cluster assignments

cluster_probability.celda_G

Calculates the conditional probability of each cell belong to each cluster given all other cluster assignments