Read and validate a contrasts file against sample metadata

Checks: 1. No duplicate contrast IDs. Ensure that the required columns (variable, reference, target) are present. 2. Values in the contrast variable column exist as column names in the sample metadata. 3. If blocking factors are supplied, checks that they are present in the sample metadata. 4. Design matrix is full rank. 5. Warn about continuous covariates (e.g. numeric patient IDs treated as continuous). 6. Values of specified columns don't contain special characters. 7. Verify that the specified reference and target values exist in the corresponding sample metadata column. 8. Issue a warning if the reference and target levels are identical.

Usage

read_contrasts(
  filename,
  samples,
  variable_column = "variable",
  reference_column = "reference",
  target_column = "target",
  blocking_column = "blocking",
  convert_to_list = FALSE,
  validate_design = TRUE
)

Arguments

filename

Contrasts file

samples

Data frame of sample information

variable_column

Column in contrasts file referencing sample sheet column

reference_column

Column in contrast file referencing reference level of sample sheet variable

target_column

Column in contrast file referencing target level of sample sheet variable

blocking_column

Colon-separated column in contrast file referencing sample sheet variables to be used as blocking factors

convert_to_list

Convert output to a list as used internally by shinyngs?

validate_design

Validate design matrix (check for NAs, full rank, numeric columns, special characters)? Set to FALSE to skip these checks.

Value

output Validated contrasts data frame

Examples

samples <- data.frame(
  condition = rep(c("control", "treated"), each = 3),
  row.names = paste0("s", 1:6)
)
contrasts_file <- tempfile(fileext = ".csv")
write.csv(
  data.frame(
    id = "treated_vs_control",
    variable = "condition",
    reference = "control",
    target = "treated"
  ),
  contrasts_file, row.names = FALSE
)
read_contrasts(contrasts_file, samples)
#>                   id  variable reference  target blocking
#> 1 treated_vs_control condition   control treated       NA