Quality Control (QC)

[MicTott]

QC roadmap

Step 1: Detect and remove empty droplets using DropletUtils::emptyDrops

• Calculate barcode ranks (barcodeRanks) and export droplet scores
• Visualize droplets scores using elbow plot

Step 2: Remove empty droplets and calculate QC metrics per droplet

• Remove empty droplets
• Ensure each species has the same mito genes
• Calculate QC metrics (sum_umi, sum_gene, mito_percent; 'scuttle::addPerCellQC')
• Detect outliers based on QC metrics (> 3 upper MAD for mito_percent, < 3 lower MAD for sum_umi/sum_gene) using scuttle:isOutlier
• Detect doublets using scDblFinder::computeDoubletDensity with 2000 HVGs; drop if doubletScore >= 2.75.

NOTE: Because one sample had a bimodal distribution for mito_percent and 2 for sum_umi/sum_gene, I subset to all other samples to calculate a shared MAD for outlier detection (built in functionality of subset input for 'scuttle::isOutlier').