CAM {CAMTHC}R Documentation

Convex Analysis of Mixtures

Description

This function performs a fully unsupervised computational deconvolution to identify marker genes that define each of the multiple subpopulations, and estimate the proportions of these subpopulations in the mixture tissues as well as their respective expression profiles.

Usage

CAM(data, K = NULL, corner.strategy = 2, dim.rdc = 10, thres.low = 0.05,
  thres.high = 0.95, cluster.method = c("K-Means", "apcluster"),
  cluster.num = 50, MG.num.thres = 20, lof.thres = 0.02, cores = NULL)

Arguments

data

Matrix of mixture expression profiles. Data frame, SummarizedExperiment or ExpressionSet object will be internally coerced into a matrix. Each row is a gene and each column is a sample. Data should be in non-log linear space with non-negative numerical values (i.e. >= 0). Missing values are not supported. All-zero rows will be removed internally.

K

The candidate subpopulation number(s), e.g. K = 2:8.

corner.strategy

The method to find corners of convex hull. 1: minimum sum of margin-of-errors; 2: minimum sum of reconstruction errors. The default is 2.

dim.rdc

Reduced data dimension; should be not less than maximum candidate K.

thres.low

The lower bound of percentage of genes to keep for CAM with ranked norm. The value should be between 0 and 1. The default is 0.05.

thres.high

The higher bound of percentage of genes to keep for CAM with ranked norm. The value should be between 0 and 1. The default is 0.95.

cluster.method

The method to do clustering. The default "K-Means" will use kmeans. The alternative "apcluster" will use apclusterK-methods.

cluster.num

The number of clusters; should be much larger than K. The default is 50.

MG.num.thres

The clusters with the gene number smaller than MG.num.thres will be treated as outliers. The default is 20.

lof.thres

Remove local outlier using lofactor. MG.num.thres is used as the number of neighbors in the calculation of the local outlier factors. The default value 0.02 will remove top 2% local outliers. Zero value will disable lof.

cores

The number of system cores for parallel computing. If not provided, one core for each element in K will be invoked. Zero value will disable parallel computing.

Details

This function includes three necessary steps to decompose a matrix of mixture expression profiles: data preprocessing, marker gene cluster search, and matrix decomposition. They are implemented in CAMPrep, CAMMGCluster and CAMASest, separately. More details can be found in the help document of each function.

For this function, you needs to specify the range of possible subpopulation numbers and the percentage of low/high-expressed genes to be removed. Typically, 30% ~ 50% low-expressed genes can be removed from gene expression data. The removal of high-expressed genes has much less impact on results, and usually set to be 0% ~ 10%.

This function can also analyze other molecular expression data, such as proteomics data. Much less low-expressed proteins need to be removed, e.g. 0% ~ 10%, due to a limited number of proteins without missing values.

Value

An object of class "CAMObj" containing the following components:

PrepResult

An object of class "CAMPrepObj" containing data preprocessing results from CAMPrep function.

MGResult

A list of "CAMMGObj" objects containing marker gene detection results from CAMMGCluster function for each K value.

ASestResult

A list of "CAMASObj" objects containing estimated proportions, subpopulation-specific expressions and mdl values from CAMASest function for each K value.

Examples

#obtain data
data(ratMix3)
data <- ratMix3$X

#set seed to generate reproducible results
set.seed(111)

#CAM with known subpopulation number
rCAM <- CAM(data, K = 3, dim.rdc = 3, thres.low = 0.30, thres.high = 0.95)
#A larger dim.rdc can improve performance but increase time complexity


#CAM with a range of subpopulation number
rCAM <- CAM(data, K = 2:5, dim.rdc = 10, thres.low = 0.30, thres.high = 0.95)

#Use "apcluster" to aggregate gene vectors in CAM
rCAM <- CAM(data, K = 2:5, dim.rdc = 10, thres.low = 0.30, thres.high = 0.95,
            cluster.method = 'apcluster')


[Package CAMTHC version 1.0.0 Index]