wrapper_ABA_RSS: The wrapper function to calculate ABA-RSS with the following...

Description Usage Arguments Value

View source: R/RSS.r

Description

This function integrates the basic ABA-RSS analysis, including RSS calculation, dimension reduction for visualization using tSNE, sample (cell) clustering based on the RSS matrix, progenitor/neuron classification with potential intermediate progenitor clusters identified.

Usage

1
2
3
4
5
6
7
wrapper_ABA_RSS(input, onlyFetal = TRUE, highVar = TRUE,
  tsneMethod = c("pca", "cor_dist"), tsne_pcaDim = 100,
  tsne_perplexity = 50, clustMethod = c("pam", "kmeans", "pca-kmeans"),
  numClust = 20, maxCells = 50000, seedset = NULL,
  progenitorScores = NULL, nameMap = NULL, progenitorModel = NULL,
  forceRetrain = TRUE, returnModel = TRUE, progenitorThres = 0.5,
  threads = 1, verbose = TRUE)

Arguments

input

The input matrix of sample (cell) transcriptome profiles. Should be properly normalized and log-transformed. Its rows representing genes should be named by ENSEMBL IDs, gene symbols or Entrez IDs.

onlyFetal

Whether only considering the fetal ABA reference samples.

highVar

Whether only considering the highly variable genes across the ABA reference samples.

tsneMethod

The method used for tSNE (Rtsne).

tsne_pcaDim

When pca method is used for tSNE, it defines the initial dimension used by PCA.

tsne_perplexity

The perplexity parameter of tSNE.

clustMethod

The method used to do cell clustering.

numClust

The expected number of clusters

maxCells

The maximum number of cells selected for clustering. NULL or NA for unlimitation.

progenitorScores

The pre-calculated progenitor scores. When it is NULL, the score calculation is involved in the pipeline.

nameMap

A matrix/data.frame showing the gene name mappings. It is expected to be of multiple columns, one of which is for gene symbols.

progenitorModel

The model used for progenitor scores estimation. It is expected to be a list with at least two components: 'gene' for the list of genes to be ranked; 'coefficients' for the model coefficients. When it is NULL, either the default or a re-trained model is used.

forceRetrain

When it is TRUE, the progenitor model is always retrained using only genes appears in the input matrix.

returnModel

Whether to return the retrained model for progenitor score estimation.

progenitorThres

The progenitor score threshold to distinct progenitors and neurons

threads

The number of threads to run.

verbose

Whether to output the progress information.

seedIdx

The indices of samples (cells) used for clustering. The remaining samples will be assigned to clusters based on the SVM-classification model.

Value

A list with all the results involved


maplesword/RefSimSpec documentation built on May 23, 2019, 1:47 p.m.