PCLasso2 {PCLassoReg}

Protein complex-based group Lasso-logistic model

Description

Protein complex-based group Lasso-logistic model

Usage

PCLasso2(
  x,
  y,
  group,
  penalty = c("grLasso", "grMCP", "grSCAD"),
  family = c("binomial", "gaussian", "poisson"),
  gamma = 8,
  standardize = TRUE,
  ...
)

Arguments

x

A n x p matrix of gene/protein expression measurements with n samples and p genes/proteins.

y

The response vector.

group

A list of groups. The feature (gene/protein) names in group should be consistent with the feature (gene/protein) names in x.

penalty

The penalty to be applied to the model. For group selection, one of grLasso, grMCP, or grSCAD. See grpreg in the R package grpreg for details.

family

Either "binomial" or "gaussian", depending on the response.

gamma

Tuning parameter of the grSCAD/grMCP penalty. Default is 8.

standardize

Logical flag for x standardization, prior to fitting the model. Default is TRUE.

...

Arguments to be passed to grpreg in the R package grpreg.

Details

The PCLasso2 model is a classification model that selects important predictors at the protein complex level to achieve accurate classification and identify risk protein complexes. The PCLasso2 model has three inputs: a protein expression matrix, a vector of binary response variables, and a number of known protein complexes. It estimates the correlation between protein expression and response variable at the level of protein complexes. Similar to traditional Lasso-logistic model, PCLasso2 is based on the logistic regression model and estimates the logistic regression coefficients by maximizing likelihood function with regularization penalty. The difference is that PCLasso2 selects features at the level of protein complexes rather than individual proteins. Considering that proteins usually function by forming protein complexes, PCLasso2 regards proteins belonging to the same protein complex as a group and constructs a group Lasso penalty (l1/l2 penalty) based on the sum (i.e. l1 norm) of the l2 norms of the regression coefficients of the group members to perform the selection of features at the group level. With the group Lasso penalty, PCLasso2 trains the logistic regression model and obtains a sparse solution at the protein complex level, that is, the proteins belonging to a protein complex are either wholly included or wholly excluded from the model. PCLasso2 outputs a prediction model and a small set of protein complexes included in the model, which are referred to as risk protein complexes. The PCSCAD and PCMCP are performed by setting the penalty parameter penalty as grSCAD and grMCP, respectively.

Value

An object with S3 class PCLasso2 containing:

fit

An object of class grpreg

Complexes.dt

Complexes with features (genes/proteins) not included in x being filtered out.

References

PCLasso2: a protein complex-based, group Lasso-logistic model for risk protein complex discovery. To be published.

PCLasso: a protein complex-based, group lasso-Cox model for accurate prognosis and risk protein complex discovery. Brief Bioinform, 2021.

Park, H., Niida, A., Miyano, S. and Imoto, S. (2015) Sparse overlapping group lasso for integrative multi-omics analysis. Journal of computational biology: a journal of computational molecular cell biology, 22, 73-84.

See Also

predict.PCLasso2, cv.PCLasso2

Examples

# load data
data(classData)
data(PCGroups)

x = classData$Exp
y = classData$Label

PC.Human <- getPCGroups(Groups = PCGroups, Organism = "Human",
Type = "GeneSymbol")

# fit PCLasso2 model
fit.PCLasso2 <- PCLasso2(x, y, group = PC.Human, penalty = "grLasso",
family = "binomial")

# fit PCSCAD model
fit.PCSCAD <- PCLasso2(x, y, group = PC.Human, penalty = "grSCAD",
family = "binomial", gamma = 10)

# fit PCMCP model
fit.PCMCP <- PCLasso2(x, y, group = PC.Human, penalty = "grMCP",
family = "binomial", gamma = 9)