| DEET_feature_extract {DEET} | R Documentation |
DEET_feature_extract
Description
Identify which genes are associated with pieces of metadata that a researcher queries.
Usage
DEET_feature_extract(mat, response, datatype, detection_cutoff = 0.7)
Arguments
mat |
A gene-by-study matrix populated by the coefficients of that study. By default, the coefficient is the log2Fold-change of genes as long as they are differentially expressed (cutoff = padj < 0.05). |
response |
A vector (binomial, categorical, or continuous) that is used to associated the DEGs within the studies. |
datatype |
indication of whether the response variable is binomial, categorical, or continuous. |
detection_cutoff |
Proportion of studies where the gene is detected (not as DE but detected at all, designated with a FC != 0). Default value 0.7. |
Value
Named list given the elastic net coefficients and the eleastic net regression between the response variable and the DEGs within DEET. It also outputs the correlation, ANOVA, and wilcoxon test of every gene against the response variable based on if it's continuous, categorical, or binomial in nature.
elastic_net_coefficients - Association that a gene has with the response variable based on the elastic net regression.
elastic_net - Output of the elastic net regression
- basic_features gives the output of the correlation, ANOVA, and wilcoxon test of every gene against the response variable.
Author(s)
Dustin Sokolowski, Jedid Ahn
References
Engebretsen, S., & Bohlin, J. (2019). Statistical predictions with glmnet. Clinical epigenetics, 11(1), 1-3.
Examples
data(DEET_feature_extract_example_matrix)
data(DEET_feature_extract_example_response)
single1 <- DEET_feature_extract(DEET_feature_extract_example_matrix,
DEET_feature_extract_example_response,"categorical")