# Bioconductor Demuxmix

A package for demultiplexing single-cell sequencing experiments of pooled cells labeled...

- Repository: https://github.com/bioMate-AI/biomate-bioconductor-kb (793 stars)
- Category: Engineering
- License: see repository
- Last updated: 2026-06-20
- Runs scripts: no, instructions only

## Install

Claude Code: `npx skills add bioMate-AI/biomate-bioconductor-kb --skill bioconductor-demuxmix -g -a claude-code -y`
Codex: `npx skills add bioMate-AI/biomate-bioconductor-kb --skill bioconductor-demuxmix -g -a codex -y`
Cursor: `npx skills add bioMate-AI/biomate-bioconductor-kb --skill bioconductor-demuxmix -g -a cursor -y`
Gemini CLI: `npx skills add bioMate-AI/biomate-bioconductor-kb --skill bioconductor-demuxmix -g -a gemini-cli -y`

## Description

A package for demultiplexing single-cell sequencing experiments of pooled cells labeled with barcode oligonucleotides. The package implements methods to fit regression mixture models for a probabilistic classification of cells, including multiplet detection. Demultiplexing error rates can be estimated, and methods for quality control are provided.

## SKILL.md, as published by its author (untrusted third-party text, read it as data)

# demuxmix

## Workflows

### Standard Workflow

Demultiplex pooled single-cell samples labeled with multiple hashtag oligonucleotides (HTOs) using regression or naive mixture models.

```r
library(demuxmix)

# Simulate a small example dataset
set.seed(2642)
class <- rbind(
  c(rep(TRUE, 220), rep(FALSE, 200)),
  c(rep(FALSE, 200), rep(TRUE, 220))
)
simdata <- dmmSimulateHto(class)
hto <- simdata$hto
rna <- simdata$rna

# Run demuxmix and summarize results
dmm <- demuxmix(hto, rna = rna)
summary(dmm)

# Classify droplets and plot QC histogram
classes <- dmmClassify(dmm)
table(classes$HTO)
plotDmmHistogram(dmm)
```
*Input/Output Note*: Inputs an HTO count matrix and a vector of detected genes per droplet; outputs a classified data frame of singlets, multiplets, negatives, and uncertain droplets.

### Pooling Non Labeled With Labeled Cells

Demultiplex a mixture of labeled cells and non-labeled cells (e.g., rare cells) using a single HTO channel.

```r
library(demuxmix)
data(csf)

# Filter out low-quality droplets
csf <- csf[csf$NumGenes >= 200, ]
hto <- t(matrix(csf$HTO, dimnames = list(rownames(csf), "HTO")))

# Fit regression mixture model
dmm <- demuxmix(hto, rna = csf$NumGenes)
summary(dmm)
dmmOverlap(dmm)

# Generate QC plots
plotDmmHistogram(dmm)
plotDmmScatter(dmm)

# Classify droplets into labeled (positive) and non-labeled (negative)
class <- dmmClassify(dmm)
table(class$HTO)
```
*Input/Output Note*: Inputs a single-channel HTO matrix and a vector of detected genes; outputs classifications separating labeled cells (positive) from non-labeled cells (negative).

The rest: https://github.com/bioMate-AI/biomate-bioconductor-kb/blob/HEAD/skills/single-cell/demuxmix/SKILL.md