In stephaniehicks/benchmarkfdrData2019: Data and Benchmarking Results from Korthauer and Kimes et al. (2019)

knitr::opts_chunk$set(echo = TRUE)

Summary

Here we download and analyze the Goodrich obesity (OB) dataset (Goodrich et al., 2014). The dataset includes fecal samples from 428 individuals with lean body mass index (BMI < 25), and 185 obese individuals (BMI > 30). We'll download the processed OTU tables from Zenodo and unzip them in the data/ob_goodrich_results folder.

We expect few or no truly differentially abundant OTUs (Finucane et al., 2014 and Walters et al., 2014), so this study provides a real-life "null" example.

Workspace Setup

library(dplyr)
library(tidyr)
library(ggplot2)
library(magrittr)
library(SummarizedBenchmark)

## load helper functions
for (f in list.files("../R", "\\.(r|R)$", full.names = TRUE)) {
    source(f)
}

## project data/results folders
datdir <- "data"
resdir <- "results"
sbdir <- "../../results/microbiome"
dir.create(datdir, showWarnings = FALSE)
dir.create(resdir, showWarnings = FALSE)
dir.create(sbdir, showWarnings = FALSE)

ob_data <- file.path(datdir, "ob_goodrich_results.tar.gz")
ob_dir <- file.path(datdir, "ob_goodrich_results")

otu_result_file <- file.path(resdir, "goodrich-otus-results.rds")
otu_bench_file <- file.path(sbdir, "goodrich-otus-benchmark.rds")
gns_result_file <- file.path(resdir, "goodrich-genus-results.rds")
gns_bench_file <- file.path(sbdir, "goodrich-genus-benchmark.rds")
gns_bench_file_abun <- file.path(sbdir, "goodrich-genus-abun-benchmark.rds")
gns_bench_file_uninf <- file.path(sbdir, "goodrich-genus-uninf-benchmark.rds")

Data Preparation

if (!file.exists(ob_data)) {
    download.file("https://zenodo.org/record/840333/files/ob_goodrich_results.tar.gz",
                  destfile = ob_data)
}

if (!file.exists(file.path(ob_dir, "ob_goodrich.metadata.txt"))) {
    untar(ob_data, exdir = datdir)
}

Next, we'll read in the unzipped OTU table and metadata files into R.

We'll need to manually define which disease labels are of interest and what metadata column contains the sample IDs (i.e. the IDs in the OTU table).

## load OTU table and metadata
otu <- read.table(file.path(ob_dir, "RDP",
                            "ob_goodrich.otu_table.100.denovo.rdp_assigned"))
meta <- read.csv(file.path(ob_dir, "ob_goodrich.metadata.txt"), sep='\t')

## Keep only samples with the right DiseaseState metadata
meta <- dplyr::filter(meta, DiseaseState %in% c("H", "OB"))

## keep only samples with both metadata and 16S data
keep_samples <- intersect(colnames(otu), meta$X)
otu <- otu[, keep_samples]
meta <- dplyr::filter(meta, X %in% keep_samples)

Since we'll be using OTU-wise covariates, we shouldn't need to perform any filtering/cleaning of the OTUs, apart from removing any that are all zeros. (This may happen after removing shallow samples, I think.) We still apply a minimum threshold of 10 reads per OTU across all samples. After removing these shallow OTUs, we also get rid of any samples with too few reads. We define the minimum number of reads per OTU in min_otu, and the minimum number of reads per sample in min_sample.

After we've removed any shallow samples, we'll convert the OTU table to relative abundances.

min_otu <- 10
minp_otu <- 0.01
min_sample <- 100

## Remove OTUs w/ <= min reads, w/ <= min prop, samples w/ <= min reads
otu <- otu[rowSums(otu) > min_otu, ]
otu <- otu[rowSums(otu > 0) / ncol(otu) > minp_otu, ]
otu <- otu[, colSums(otu) > min_sample]

# Update metadata with new samples
meta <- dplyr::filter(meta, X %in% colnames(otu))

# Remove empty OTUs
otu <- otu[rowSums(otu) > 0, ]

# Convert to relative abundance
abun_otu <- t(t(otu) / rowSums(t(otu)))

# Add pseudo counts
zeroabun <- 0
abun_otu <- abun_otu + zeroabun

Data Analysis

Differential Testing

Next, we need to calculate the pvalues, effect size, and standard error for each OTU. Here, we'll compare lean vs. obese. We'll put these results into a dataframe, and label the columns with the standardized names for downstream use (pval, SE, effect_size, test_statistic). The test statistic is the one returned by wilcox.test().

Note that the effect here is calculated as logfold change of mean abundance in controls relative to cases (i.e. log(mean_abun[controls]/mean_abun[cases])).

While we're at it, we'll also calculate the mean abundance and ubiquity (detection rate) of each OTU. Later, we can assign their values to a new column called ind_covariate for use in downstream steps.

Note that OB Goodrich has ~70,000 OTUs, so this is going to take a while.

if (!file.exists(otu_result_file)) {
    res <- test_microbiome(abundance = abun_otu, shift = zeroabun,
                           is_case = meta$DiseaseState == "OB")
    saveRDS(res, file = otu_result_file)
} else {
    res <- readRDS(otu_result_file)
}

Finally, let's try to add phylogeny as covariates. Here we'll have columns for each separate taxonomic level.

res <- tidyr::separate(res, otu, 
                       c("kingdom", "phylum", "class", "order",
                         "family", "genus", "species", "denovo"), 
                       sep=";", remove = FALSE)

Covariate Diagnostics

Here we look to see if the covariates do indeed look informative.

Ubiquity

strat_hist(res, pvalue="pval", covariate="ubiquity", maxy=5, binwidth=0.05, numQ=4)

rank_scatter(res, pvalue="pval", covariate="ubiquity")

Mean Abundance (across non-zero samples)

strat_hist(res, pvalue="pval", covariate="mean_abun_present", maxy=5, binwidth=0.05, numQ=4)

rank_scatter(res, pvalue="pval", covariate="mean_abun_present")

Phylogeny

Let's look at phylum-level stratification first. A priori, I might expect Proteobacteria to be enriched for low p-values? But I don't know if that's super legit, and Eric doesn't seem to think that phylogeny will be informative at all...

ggplot(res, aes(x=pval)) +
    geom_histogram() +
    facet_wrap(~phylum, scales = "free")

Multiple-Testing Correction - ubiquity

Let's use ubiquity as our ind_covariate.

res <- dplyr::mutate(res, ind_covariate = ubiquity)

First, we'll create an object of BenchDesign class to hold the data and add the benchmark methods to the BenchDesign object. We remove ASH from the default comparison.

Next, we'll construct the SummarizedBenchmark object, which will run the functions specified in each method (these are actually sourced in from the helper scripts).

if (!file.exists(otu_bench_file)) {
    bd <- initializeBenchDesign()
    bd <- dropBMethod(bd, "ashq")
    sb <- buildBench(bd, data = res, ftCols = "ind_covariate")
    metadata(sb)$data_download_link <-
                   "https://zenodo.org/record/840333/files/ob_goodrich_results.tar.gz"
    saveRDS(sb, file = otu_bench_file)
} else {
    sb <- readRDS(otu_bench_file)
}

Benchmark Metrics - ubiquity

Next, we'll add the default performance metric for q-value assays. First, we have to rename the assay to 'qvalue'.

assayNames(sb) <- "qvalue"
sb <- addDefaultMetrics(sb)

Now, we'll plot the results.

## plot nrejects by method overall and stratified by covariate
rejections_scatter( sb, as_fraction=FALSE, supplementary=FALSE)
rejection_scatter_bins(sb, covariate="ind_covariate", supplementary=FALSE)
##plotFDRMethodsOverlap(sb, alpha=0.1, supplementary=FALSE )

covariateLinePlot(sb, alpha = 0.05, covname = "ind_covariate")

There is no upsetR plot for OB Goodrich for now because only lfdr is returning any rejections.

assays(sb)[["qvalue"]][, "ashq"] %>% max()
sum(is.na(assays(sb)[["qvalue"]][, "scott-empirical"]))
sum(is.na(assays(sb)[["qvalue"]][, "lfdr"]))

Data Analysis (genus-level)

Collapse to Genus

## add column with otu names
genus_df <- as.data.frame(abun_otu)
genus_df <- dplyr::as_tibble(genus_df, rownames = "otu")

## just get genus information
genus_df <- dplyr::mutate(genus_df, genus = sapply(strsplit(otu, ";"), `[`, 6))

## gather into tidy format
genus_df <- tidyr::gather(genus_df, key = "sample", value = "abun", -otu, -genus)

## get rid of unannoated genera, and sum abundances for genera
genus_df <- genus_df %>% 
    dplyr::filter(genus != "g__") %>% 
    dplyr::group_by(genus, sample) %>% 
    dplyr::summarise(total_abun = sum(abun)) %>%
    ungroup()

## convert back to longform
genus_df <- tidyr::spread(genus_df, sample, total_abun)
genus_df <- as.data.frame(as.list(dplyr::select(genus_df, -genus)),
                          row.names = genus_df$genus)

## re-order columns to match metadata
genus_df <- genus_df[, match(meta$X, colnames(genus_df))]

## use matrix
genus <- as.matrix(genus_df)

Differential Testing

if (!file.exists(gns_result_file)) {
    res <- test_microbiome(abundance = genus, shift = zeroabun,
                           is_case = meta$DiseaseState == "OB")
    saveRDS(res, file = gns_result_file)
} else {
    res <- readRDS(gns_result_file)
}

Add random (uninformative) covariate.

set.seed(76291)
res$rand_covar <- rnorm(nrow(res))

Covariate Diagnostics

Here we look to see if the covariates do indeed look informative.

Ubiquity

strat_hist(res, pvalue="pval", covariate="ubiquity", maxy=10, binwidth=0.05)

rank_scatter(res, pvalue="pval", covariate="ubiquity")

Something weird is happening with p=0.20 and p=0.40 - maybe this has something to do with ties? Either way, ubiquity looks to be a bit informative (from the scatter plot, but not really the histograms...)

Mean Abundance (across non-zero samples)

strat_hist(res, pvalue="pval", covariate="mean_abun_present", maxy=8, binwidth=0.05)

rank_scatter(res, pvalue="pval", covariate="mean_abun_present")

Random (uninformative)

strat_hist(res, pvalue="pval", covariate="rand_covar", maxy=5, binwidth=0.05, numQ=4)

rank_scatter(res, pvalue="pval", covariate="rand_covar")

Multiple-Testing Correction - ubiquity

Let's use ubiquity as our ind_covariate.

res <- dplyr::mutate(res, ind_covariate = ubiquity)

First, we'll create an object of BenchDesign class to hold the data and add the benchmark methods to the BenchDesign object.

Then, we'll construct the SummarizedBenchmark object, which will run the functions specified in each method (these are actually sourced in from the helper scripts).

if (!file.exists(gns_bench_file)) {
    bd <- initializeBenchDesign()
    bd <- dropBMethod(bd, "ashq")
    sb <- buildBench(bd, data = res, ftCols = "ind_covariate")
    metadata(sb)$data_download_link <-
                   "https://zenodo.org/record/840333/files/ob_goodrich_results.tar.gz"
    saveRDS(sb, file = gns_bench_file)
} else {
    sb <- readRDS(gns_bench_file)
}

Benchmark Metrics - ubiquity

Next, we'll add the default performance metric for q-value assays. First, we have to rename the assay to 'qvalue'.

assayNames(sb) <- "qvalue"
sb <- addDefaultMetrics(sb)

Now, we'll plot the results.

rejections_scatter(sb, as_fraction=FALSE, supplementary=FALSE)
rejection_scatter_bins(sb, covariate="ind_covariate", supplementary=FALSE)

covariateLinePlot(sb, alpha = 0.05, covname = "ind_covariate")

Note: Benjamini-Hochberg (bh) overlaps exactly with the IHW results.

plotFDRMethodsOverlap(sb, alpha=0.1, supplementary=FALSE, order.by="freq", nsets=100 )

methods <- c( "lfdr", "ihw-a10", "bl-df03", "qvalue", "bh", "bonf" )
plotCovariateBoxplots(sb, alpha=0.1, nsets=6, methods=methods)

Multiple-Testing Correction - mean abun

Let's use mean_abun_present as our ind_covariate.

res <- dplyr::mutate(res, ind_covariate = mean_abun_present)

First, we'll create an object of BenchDesign class to hold the data and add the benchmark methods to the BenchDesign object.

Then, we'll construct the SummarizedBenchmark object, which will run the functions specified in each method (these are actually sourced in from the helper scripts).

if (!file.exists(gns_bench_file_abun)) {
    bd <- initializeBenchDesign()
    bd <- dropBMethod(bd, "ashq")
    sb <- buildBench(bd, data = res, ftCols = "ind_covariate")
    metadata(sb)$data_download_link <-
                   "https://zenodo.org/record/840333/files/ob_goodrich_results.tar.gz"
    saveRDS(sb, file = gns_bench_file_abun)
} else {
    sb <- readRDS(gns_bench_file_abun)
}

Benchmark Metrics - mean abun

Next, we'll add the default performance metric for q-value assays. First, we have to rename the assay to 'qvalue'.

assayNames(sb) <- "qvalue"
sb <- addDefaultMetrics(sb)

Now, we'll plot the results.

rejections_scatter(sb, as_fraction=FALSE, supplementary=FALSE)
rejection_scatter_bins(sb, covariate="ind_covariate", supplementary=FALSE)

covariateLinePlot(sb, alpha = 0.05, covname = "ind_covariate")

Note: Benjamini-Hochberg (bh) overlaps exactly with the IHW results.

plotFDRMethodsOverlap(sb, alpha=0.1, supplementary=FALSE, order.by="freq", nsets=100 )

methods <- c( "lfdr", "ihw-a10", "bl-df03", "qvalue", "bh", "bonf" )
plotCovariateBoxplots(sb, alpha=0.1, nsets=6, methods=methods)

Multiple-Testing Correction - random

Let's use rand_covar as our ind_covariate.

res <- dplyr::mutate(res, ind_covariate = rand_covar)

First, we'll create an object of BenchDesign class to hold the data and add the benchmark methods to the BenchDesign object.

Then, we'll construct the SummarizedBenchmark object, which will run the functions specified in each method (these are actually sourced in from the helper scripts).

if (!file.exists(gns_bench_file_uninf)) {
    bd <- initializeBenchDesign()
    bd <- dropBMethod(bd, "ashq")
    sb <- buildBench(bd, data = res, ftCols = "ind_covariate")
    metadata(sb)$data_download_link <-
                   "https://zenodo.org/record/840333/files/ob_goodrich_results.tar.gz"
    saveRDS(sb, file = gns_bench_file_uninf)
} else {
    sb <- readRDS(gns_bench_file_uninf)
}

Benchmark Metrics - random

Next, we'll add the default performance metric for q-value assays. First, we have to rename the assay to 'qvalue'.

assayNames(sb) <- "qvalue"
sb <- addDefaultMetrics(sb)

Now, we'll plot the results.

rejections_scatter(sb, as_fraction=FALSE, supplementary=FALSE)
rejection_scatter_bins(sb, covariate="ind_covariate", supplementary=FALSE)

covariateLinePlot(sb, alpha = 0.05, covname = "ind_covariate")

Note: Benjamini-Hochberg (bh) overlaps exactly with the IHW results.

plotFDRMethodsOverlap(sb, alpha=0.1, supplementary=FALSE, order.by="freq", nsets=100 )

methods <- c( "lfdr", "ihw-a10", "bl-df03", "qvalue", "bh", "bonf" )
plotCovariateBoxplots(sb, alpha=0.1, nsets=6, methods=methods)

OTU/Genus comparison

Here we compare the method ranks for the different comparisons at alpha = 0.10.

plotMethodRanks(c(otu_bench_file, gns_bench_file, gns_bench_file_abun, gns_bench_file_uninf), 
                colLabels = c("OTU", "Genus-ubiquity", "Genus-abun", "Genus-uninf"), 
                alpha = 0.10, xlab = "Comparison")

Session Info

sessionInfo()

stephaniehicks/benchmarkfdrData2019 documentation built on June 20, 2021, 10 a.m.