BASiCS_Chain: The BASiCS_Chain class

Description Slots Author(s) Examples

Description

Container of an MCMC sample of the BASiCS' model parameters as generated by the function BASiCS_MCMC.

Slots

parameters

List of matrices containing MCMC chains for each model parameter. Depending on the mode in which BASiCS was run, the following parameters can appear in the list:

mu

MCMC chain for gene-specific mean expression parameters μ_i, biological genes only (matrix with q.bio columns, all elements must be positive numbers)

delta

MCMC chain for gene-specific biological over-dispersion parameters δ_i, biological genes only (matrix with q.bio columns, all elements must be positive numbers)

phi

MCMC chain for cell-specific mRNA content normalisation parameters φ_j (matrix with n columns, all elements must be positive numbers and the sum of its elements must be equal to n)

This parameter is only used when spike-in genes are available.

s

MCMC chain for cell-specific technical normalisation parameters s_j (matrix with n columns, all elements must be positive numbers)

nu

MCMC chain for cell-specific random effects ν_j (matrix with n columns, all elements must be positive numbers)

theta

MCMC chain for technical over-dispersion parameter(s) θ (matrix, all elements must be positive, each colum represents 1 batch)

beta

Only relevant for regression BASiCS model (Eling et al, 2017). MCMC chain for regression coefficients (matrix with k columns, where k represent the number of chosen basis functions + 2)

sigma2

Only relevant for regression BASiCS model (Eling et al, 2017). MCMC chain for the residual variance (matrix with one column, sigma2 represents a global parameter)

epsilon

Only relevant for regression BASiCS model (Eling et al, 2017). MCMC chain for the gene-specific residual over-dispersion parameter (matrix with q columns)

RefFreq

Only relevant for no-spikes BASiCS model (Eling et al, 2017). For each biological gene, this vector displays the proportion of times for which each gene was used as a reference (within the MCMC algorithm), when using the stochastic reference choice described in (Eling et al, 2017). This information has been kept as it is useful for the developers of this library. However, we do not expect users to need it.

Author(s)

Catalina A. Vallejos cnvallej@uc.cl

Nils Eling eling@ebi.ac.uk

Examples

1
2
3
4
5
6
7
8
# A BASiCS_Chain object created by the BASiCS_MCMC function.
Data <- makeExampleBASiCS_Data()

# To run the model without regression
Chain <- BASiCS_MCMC(Data, N = 100, Thin = 2, Burn = 2, Regression = FALSE)

# To run the model using the regression model
ChainReg <- BASiCS_MCMC(Data, N = 100, Thin = 2, Burn = 2, Regression = TRUE)

BASiCS documentation built on April 16, 2021, 6 p.m.