HRSIP {HTSSIP}

(MW-)HR-SIP analysis

Description

Conduct (multi-window) high resolution stable isotope probing (HR-SIP) analysis.

Usage

HRSIP(physeq, design, density_windows = data.frame(density_min = c(1.7),
  density_max = c(1.75)), sparsity_threshold = seq(0, 0.3, 0.1),
  sparsity_apply = "all", l2fc_threshold = 0.25, padj_method = "BH",
  padj_cutoff = NULL, parallel = FALSE)

Arguments

physeq

Phyloseq object

design

design parameter used for DESeq2 analysis. This is usually used to differentiate labeled-treatment and unlabeld-control samples. See DESeq2::DESeq for more details on the option.

density_windows

The buoyant density window(s) used for for calculating log2 fold change values. Input can be a vector (length 2) or a data.frame with a 'density_min' and a 'density_max' column (each row designates a density window).

sparsity_threshold

All OTUs observed in less than this portion (fraction: 0-1) of gradient fraction samples are pruned. This is a form of indepedent filtering. The sparsity cutoff with the most rejected hypotheses is used.

sparsity_apply

Apply sparsity threshold to all gradient fraction samples ('all') or just 'heavy' fraction samples ('heavy'), where 'heavy' samples are designated by the density_windows.

l2fc_threshold

log2 fold change (l2fc) values must be significantly above this threshold in order to reject the hypothesis of equal counts. See DESeq2 for more information.

padj_method

Method for global p-value adjustment (See p.adjust()).

padj_cutoff

Adjusted p-value cutoff for rejecting the null hypothesis that l2fc values were not greater than the l2fc_threshold. Set to NULL to skip filtering of results to the sparsity cutoff with most rejected hypotheses and filtering each OTU to the buoyant density window with the greatest log2 fold change.

parallel

Process each parameter combination in parallel. See plyr::mdply() for more information.

Details

The (MW-)HR-SIP workflow is as follows:

1. For each sparsity threshold & BD window: calculate log2 fold change values (with DESeq2) for each OTU

2. Globally adjust p-values with a user-defined method (see p.adjust())

3. Select the sparsity cutoff with the most rejected hypotheses

4. For each OTU, select the BD window with the greatest log2 fold change value

Value

dataframe of HRSIP results

Examples

data(physeq_S2D2_l)

## Not run: 
# HR-SIP on just 1 treatment-control comparison
## 1st item in list of phyloseq objects
physeq = physeq_S2D2_l[[1]]
## HR-SIP
### Note: treatment-control samples differentiated with 'design=~Substrate'
df_l2fc = HRSIP(physeq, design=~Substrate)
head(df_l2fc)

## Same, but multiple BD windows (MW-HR-SIP) & run in parallel
### Windows = 1.7-1.73 & 1.72-1.75
doParallel::registerDoParallel(2)
dw = data.frame(density_min=c(1.7, 1.72), density_max=c(1.73, 1.75))
df_l2fc = HRSIP(physeq_S2D1_l[[1]],
                design=~Substrate,
                density_windows=dw,
                parallel=TRUE)
head(df_l2fc)

## End(Not run)