Skip to contents

Remove negative-control contaminants from a phyloseq object

Source: R/negative_controls.R
neg_control_clean_pq.Rd

lifecycle-experimental

Drop every taxon that is detected above a read threshold in negative controls, then optionally drop the negative-control samples themselves.

This is a simpler, more aggressive alternative to decontam_sam_control(): instead of zeroing OTU values at or below the per-taxon control threshold, it removes the offending taxa altogether. Use it when any presence in a negative control is considered disqualifying.

Algorithm:

  1. Subset physeq to its negative-control samples.

  2. Identify taxa whose summed reads across those samples is strictly greater than min_reads.

  3. Remove those taxa from the full dataset.

  4. If drop_neg_samples = TRUE (default), also drop the negative-control samples.

Usage

neg_control_clean_pq(
  physeq,
  neg_control,
  min_reads = 2,
  drop_neg_samples = TRUE,
  clean_phyloseq_object = TRUE,
  verbose = TRUE
)

Arguments

physeq

(phyloseq, required) A phyloseq object.

neg_control

An expression evaluated on sample_data that returns TRUE for negative-control samples (e.g., is_control == TRUE, sample_type == "blank"). Use . to refer to the phyloseq object.

min_reads

(integer, default: 2) Read-count threshold for a taxon to count as a contaminant. A taxon is removed when its total reads across NC samples is strictly greater than min_reads. The default 2 matches the common rule "tolerate up to two stray reads in NCs".

drop_neg_samples

(logical, default: TRUE) Drop the negative-control samples from the returned object. Set to FALSE to keep them (e.g., for NC-aware downstream modelling).

clean_phyloseq_object

(logical, default: TRUE) Whether to clean the resulting phyloseq object using MiscMetabar::clean_pq() to remove empty taxa/samples.

verbose

(logical, default: TRUE) Whether to print a short message reporting the number of contaminant taxa removed.

Value

A phyloseq object with contaminant taxa (and, by default, NC samples) removed.

Author

Adrien Taudière

Examples

library(MiscMetabar)
data(data_fungi)
# Mark the three lowest-depth samples as mock controls for demo
pq <- mutate_samdata_pq(
  data_fungi,
  is_control = sample_sums(.) < sort(sample_sums(.))[4]
)
neg_control_clean_pq(pq, is_control)
#> Removing 2 contaminant taxa (>2 reads in NCs).
#> phyloseq-class experiment-level object
#> otu_table()   OTU Table:         [ 1418 taxa and 182 samples ]
#> sample_data() Sample Data:       [ 182 samples by 8 sample variables ]
#> tax_table()   Taxonomy Table:    [ 1418 taxa by 12 taxonomic ranks ]
#> refseq()      DNAStringSet:      [ 1418 reference sequences ]

# Keep NC samples in output
neg_control_clean_pq(pq, is_control, drop_neg_samples = FALSE)
#> Removing 2 contaminant taxa (>2 reads in NCs).
#> phyloseq-class experiment-level object
#> otu_table()   OTU Table:         [ 1418 taxa and 185 samples ]
#> sample_data() Sample Data:       [ 185 samples by 8 sample variables ]
#> tax_table()   Taxonomy Table:    [ 1418 taxa by 12 taxonomic ranks ]
#> refseq()      DNAStringSet:      [ 1418 reference sequences ]

# Stricter: drop any taxon ever detected in NCs
neg_control_clean_pq(pq, is_control, min_reads = 0)
#> Removing 14 contaminant taxa (>0 reads in NCs).
#> phyloseq-class experiment-level object
#> otu_table()   OTU Table:         [ 1406 taxa and 182 samples ]
#> sample_data() Sample Data:       [ 182 samples by 8 sample variables ]
#> tax_table()   Taxonomy Table:    [ 1406 taxa by 12 taxonomic ranks ]
#> refseq()      DNAStringSet:      [ 1406 reference sequences ]