system_genetics/rnaseq/step6_overall_QC/03 - Sample Counts.R

# Total counts per sample
# Normalized with limma::voom

source("__ - Preloader.R", verbose=T)


# The analysis
norm.expr.data <- expression.data %>%
	tibble::column_to_rownames("Gene")
norm.expr.data <- norm.expr.data[rowSums(norm.expr.data) >= 10,] %>%
	limma::voom() %>%
	as.matrix()

# Total counts per sample
total.count.per.sample <- expression.data %>%
	tibble::column_to_rownames("Gene") %>%
	colSums()

data.frame(
		sample = names(total.count.per.sample),
		counts = as.numeric(total.count.per.sample)
	) %>%
	readr::write_csv(file.path(results.dir, "total.counts.per.sample.csv"))


norm.data <- norm.expr.data %>%
  as.data.frame() %>%
  tibble::rownames_to_column(
    "Gene"
  ) %>%
  tidyr::gather(
    key = "sample.id",
    value = "expr.value",
    -Gene
  ) %>%
  dplyr::left_join(
    y = master.Table %>%
      dplyr::filter(
          !is.na(GenomeScan_ID)
      ) %>%
      dplyr::mutate(
          id = dplyr::case_when(
            stringr::str_trim(gender) == "" ~ paste0("Water ", dplyr::row_number()),
            TRUE ~ sample.id
          ),
          gender = dplyr::case_when(
            stringr::str_trim(gender) == "" ~ "water",
            !is.na(gender) ~ as.character(gender)
          )
      ) %>%
      dplyr::select(
        GenomeScan_ID,
        gender,
        id
      ),
    by = c("sample.id" = "GenomeScan_ID")
  )

norm.plot <- norm.data %>%
  ggplot2::ggplot(
    mapping = ggplot2::aes(
      x = id,
      y = expr.value,
      fill = gender
    )
  ) +
  ggplot2::geom_boxplot() +
  ggplot2::scale_fill_manual(
    values = c(
      "male" = "blue",
      "female" = "red",
      "water" = "green"
    )
  ) +
  ggplot2::labs(
    title = "Normalized expression values distribution",
    y = "Normalized expression values (limma::voom)",
    x = "Sample",
    gender = "Gender"
  ) +
  ggprism::theme_prism() +
  ggplot2::theme(
    axis.text.x = ggplot2::element_text(angle = 90)
  )

ggplot2::ggsave(
  filename = file.path(results.dir, "counts.per.sample.normalised.png"),
  plot = norm.plot,
  width = 40,
  height = 20,
  units = "cm"
)



expr.data <- expression.data %>%
  tidyr::gather(
    key = "sample.id",
    value = "expr.value",
    -Gene
  ) %>%
  dplyr::filter(
    expr.value != 0
  ) %>%
  dplyr::left_join(
    y = master.Table %>%
      dplyr::filter(
          !is.na(GenomeScan_ID)
      ) %>%
      dplyr::mutate(
          id = dplyr::case_when(
            stringr::str_trim(gender) == "" ~ paste0("Water ", dplyr::row_number()),
            TRUE ~ sample.id
          ),
          gender = dplyr::case_when(
            stringr::str_trim(gender) == "" ~ "water",
            !is.na(gender) ~ as.character(gender)
          )
      ) %>%
      dplyr::select(
        GenomeScan_ID,
        gender,
        id
      ),
    by = c("sample.id" = "GenomeScan_ID")
  )

expr.plot <- expr.data %>%
  ggplot2::ggplot(
    mapping = ggplot2::aes(
      x = id,
      y = expr.value,
      fill = gender
    )
  ) +
  ggplot2::geom_boxplot() +
  ggplot2::scale_fill_manual(
    values = c(
      "male" = "blue",
      "female" = "red",
      "water" = "green"
    )
  ) +
  ggplot2::scale_y_continuous(trans='log2')  +
  ggplot2::labs(
    title = "Raw expression values distribution, without zero's",
    y = "Expression values",
    x = "Sample",
    gender = "Gender"
  ) +
  ggprism::theme_prism() +
  ggplot2::theme(
    axis.text.x = ggplot2::element_text(angle = 90)
  )

ggplot2::ggsave(
  filename = file.path(results.dir, "counts.per.sample.raw.zeros.removed.png"),
  plot = expr.plot,
  width = 40,
  height = 20,
  units = "cm"
)