Skip to content

Add zero imputation and use lmer #34

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Open
lmartinezgili wants to merge 2 commits into
base: master
Choose a base branch
from
Open

Add zero imputation and use lmer #34

Changes from all commits
Commits
Show all changes
2 commits
Select commit Hold shift + click to select a range
2a91aba
Add zero imputation and use lmer
lmartinezgili Jul 4, 2021
44041ff
use default anova type
lmartinezgili Nov 15, 2021
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
39 changes: 30 additions & 9 deletions scripts/ancom_v2.1.R
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -2,6 +2,8 @@ library(nlme)
library(dplyr)
library(ggplot2)
library(compositions)
library(zCompositions)
library(lmerTest)

# OTU table should be a matrix/data.frame with each feature in rows and sample in columns.
# Metadata should be a matrix/data.frame containing the sample identifier.
Expand Down Expand Up @@ -121,8 +123,8 @@ feature_table_pre_process = function(feature_table, meta_data, sample_var, group
}

# ANCOM main function
ANCOM = function(feature_table, meta_data, struc_zero = NULL, main_var, p_adj_method = "BH",
alpha = 0.05, adj_formula = NULL, rand_formula = NULL, ...){
ancom = function(feature_table, meta_data, struc_zero = NULL, main_var, p_adj_method = "BH",
alpha = 0.05, adj_formula = NULL, rand_formula = NULL, imp_pseudocount, ...){
# OTU table transformation:
# (1) Discard taxa with structural zeros (if any); (2) Add pseudocount (1) and take logarithm.
if (!is.null(struc_zero)) {
Expand All @@ -131,7 +133,27 @@ ANCOM = function(feature_table, meta_data, struc_zero = NULL, main_var, p_adj_me
}else{
comp_table = feature_table
}
comp_table = log(as.matrix(comp_table) + 1)
if (imp_pseudocount== TRUE) {
comp_table = log(as.matrix(comp_table) + 1) #previously default choice
} else {
comp_table_NA= t(comp_table) #sample in rows for imputing zeros
comp_table_temp= t(comp_table)
comp_table_temp[is.na(comp_table_temp)]= 0 #temporarily substitue NA by zero for cmultRepl()

comp_table_temp_test= apply(comp_table_temp, 2, function(x) sum(x==0, na.rm=T))
if (length(which(comp_table_temp_test >= nrow(comp_table_temp)-1)) != 0) {
comp_table_temp= comp_table_temp[,-c(which(comp_table_temp_test >= nrow(comp_table_temp)-1))]
comp_table_NA= comp_table_NA[,-c(which(comp_table_temp_test >= nrow(comp_table_temp)-1))]
} else {
comp_table_temp= comp_table_temp
comp_table_NA= comp_table_NA
}
#Impute zeros
comp_table= cmultRepl(comp_table_temp, label= 0, method= "GBM", delta= 0.65, output= "p-counts")
comp_table[is.na(comp_table_NA)]= NA #re-introduce NA where outlier zeros were detected
comp_table= t(comp_table) #OTU/ASV in rows again for ancom()
comp_table= log(as.matrix(comp_table))
}
n_taxa = dim(comp_table)[1]
taxa_id = rownames(comp_table)
n_samp = dim(comp_table)[2]
Expand All @@ -156,14 +178,14 @@ ANCOM = function(feature_table, meta_data, struc_zero = NULL, main_var, p_adj_me
tformula = formula(paste("x ~", main_var, "+", adj_formula, sep = " "))
}else if (!is.null(rand_formula)) {
# Model: Mixed-effects model
tfun = nlme::lme
tfun= lmerTest::lmer
# Formula
if (is.null(adj_formula)) {
# Random intercept model
tformula = formula(paste("x ~", main_var))
tformula = paste("x ~", main_var)
}else {
# Random coefficients/slope model
tformula = formula(paste("x ~", main_var, "+", adj_formula))
tformula = paste("x ~", main_var, "+", adj_formula)
}
}

Expand Down Expand Up @@ -203,11 +225,10 @@ ANCOM = function(feature_table, meta_data, struc_zero = NULL, main_var, p_adj_me
)
}else if (!is.null(rand_formula)) {
p_data[-(1:i), i] = apply(alr_data[, 1:n_lr, drop = FALSE], 2, function(x){
fit = tfun(fixed = tformula,
fit = tfun(formula(paste(tformula,"+",rand_formula)),
data = data.frame(x, alr_data, check.names = FALSE),
random = formula(rand_formula),
na.action = na.omit, ...)
anova(fit)[main_var, "p-value"]
anova(fit)[main_var, "Pr(>F)"]
}
)
}
Expand Down