In AlexsLemonade/scpcaTools: Single cell data tools for ScPCA

Multiplexing Experiment Summary

Multiplexing Statistics

# add rowData if missing
if (is.null(rowData(multiplex_exp)$detected)) {
  multiplex_exp <- scuttle::addPerFeatureQCMetrics(multiplex_exp)
}

cell_hto_counts <- Matrix::colSums(counts(multiplex_exp))

hto_information <- tibble::tibble(
  "Number of HTOs assayed" =
    format(nrow(multiplex_exp), big.mark = ",", scientific = FALSE),
  "Multiplex reads sequenced" =
    format(multiplex_meta$total_reads, big.mark = ",", scientific = FALSE),
  "Percent multiplex reads mapped to ADTs" =
    paste0(round(multiplex_meta$mapped_reads / multiplex_meta$total_reads * 100, digits = 2), "%"),
  "Percent of HTOs in cells" =
    paste0(round(sum(cell_hto_counts) / multiplex_meta$mapped_reads * 100, digits = 2), "%"),
  "Percent of cells with HTOs" =
    paste0(round(sum(cell_hto_counts > 0) / length(cell_hto_counts) * 100, digits = 2), "%"),
  "Median HTO UMIs per cell" =
    format(median(cell_hto_counts), big.mark = ",", scientific = FALSE)
) |>
  t()

knitr::kable(hto_information, align = "r") |>
  kableExtra::kable_styling(
    bootstrap_options = "striped",
    full_width = FALSE,
    position = "left"
  ) |>
  kableExtra::column_spec(2, monospace = TRUE)

Hashtag Oligos (HTOs) Statistics

hto_tags <- as.data.frame(rowData(multiplex_exp)) |>
  tibble::rownames_to_column("Hashtag Oligo (HTO)") |>
  arrange(desc(mean)) |>
  select("Hashtag Oligo (HTO)",
    "Mean UMI count per cell" = mean,
    "Percent of cells detected" = detected
  )

knitr::kable(hto_tags, digits = 2) |>
  kableExtra::kable_styling(
    bootstrap_options = c("striped", "condensed"),
    full_width = FALSE,
    position = "left",
  ) |>
  kableExtra::column_spec(2:3, monospace = TRUE)

Demultiplexing Sample Calls

Demultiplexing was performed using both DropletUtils::hashedDrops and Seurat::HTOdemux only on the filtered cells, using default parameters for each.

For multiplex libraries where bulk RNA-seq data is available for the individual samples, we also performed demultiplexing analysis using genotype data following the methods described in Weber et al. (2021). The genetic demultiplexing results are reported under the vireo column in the below table.

Note: We have reported the demultiplexed sample calls for each of the above mentioned algorithms, but the multiplexed library has not been separated into individual samples.

# grab columns that have demuxing results as not all methods could be present
# e.g. vireo is only used if a matching bulk RNA seq library is there,
# otherwise those calls will not be present
demux_methods <- c("hashedDrops_sampleid", "HTODemux_sampleid", "vireo_sampleid")
demux_columns <- colnames(colData(filtered_sce)) %in% demux_methods

if (any(demux_columns)) {
  # create a table summarizing demuxing calls for each method used
  demux_calls <- as.data.frame(colData(filtered_sce)[, demux_columns]) |>
    # remove _sampleid at the end of the column names
    dplyr::rename_with(~ stringr::str_remove(., "_sampleid")) |>
    tidyr::pivot_longer(
      cols = everything(),
      names_to = "demux_method",
      values_to = "Sample"
    ) |>
    dplyr::count(Sample, demux_method) |>
    tidyr::pivot_wider(
      names_from = demux_method,
      values_from = n
    )


  knitr::kable(demux_calls, digits = 2) |>
    kableExtra::kable_styling(
      bootstrap_options = c("striped", "condensed"),
      full_width = FALSE,
      position = "left",
    ) |>
    kableExtra::column_spec(2:ncol(demux_calls), monospace = TRUE)
} else {
  glue::glue("
    <div class=\"alert alert-info\">

    Demultiplexing was not applied to this library using any of the above mentioned methods.
    Sample calls cannot be reported.

    </div>
  ")
}