Skip to contents

Counts the genes that reach each threshold in enough samples, for a grid of thresholds and sample requirements, so that the effect of a threshold can be seen before rsem_filter() removes genes. Nothing is removed.

Usage

rsem_detection_curve(x, expr, thresholds = NULL, min_samples = c(1, 0.2, 0.5))

Arguments

x

The gene by sample table, as returned by rsem_read(): gene names as row names (or in the first column) and one numeric column per sample.

expr

"count" or "tpm": what the values are. Required.

thresholds

Values to try. Default c(0.1, 0.5, 1, 2, 5, 10) for TPM and c(1, 5, 10, 20, 50, 100) for counts.

min_samples

Sample requirements to try; a number below 1 is a proportion of the samples, as in rsem_filter(). Default c(1, 0.2, 0.5): at least one sample, 20% and half of the samples.

Value

A data frame with one row per threshold and sample requirement: threshold, min_samples (as given), samples_needed, genes (genes kept), share (of all genes in x) and genes_mean (genes whose mean reaches the threshold; the same in every row of a threshold).

Details

For a threshold \(\tau\) and a required number of samples \(k\), the count is the number of genes with a value of at least \(\tau\) in at least \(k\) samples. This is exactly the rule of rsem_filter(method = "detected", min_value = tau, min_samples = k), so each row tells how many genes that call would keep. The column genes_mean gives the number of genes whose mean over samples reaches the threshold, the other rule often used in published scripts.

Examples

if (requireNamespace("org.Hs.eg.db", quietly = TRUE)) {
  ex <- system.file("extdata", "example", package = "rsemverse")
  tpm <- rsem_read(file.path(ex, "data"), file.path(ex, "metadata.csv"),
                   species = "human", expr = "tpm")
  rsem_detection_curve(tpm, expr = "tpm")
}
#> 
#> rsem_read: 3 samples, 44 genes -> 43 rows (species = human, expr = tpm, org.Hs.eg.db 3.23.1)
#>   with symbol: 42 genes (1 symbols merged from several IDs); kept as original IDs: 2 genes
#> rsem_detection_curve: 43 genes x 3 samples; 6 thresholds x 3 sample requirements (TPM)
#>    threshold min_samples samples_needed genes share genes_mean
#> 1        0.1         1.0              1    43     1         43
#> 2        0.1         0.2              1    43     1         43
#> 3        0.1         0.5              2    43     1         43
#> 4        0.5         1.0              1    43     1         43
#> 5        0.5         0.2              1    43     1         43
#> 6        0.5         0.5              2    43     1         43
#> 7        1.0         1.0              1    43     1         43
#> 8        1.0         0.2              1    43     1         43
#> 9        1.0         0.5              2    43     1         43
#> 10       2.0         1.0              1    43     1         43
#> 11       2.0         0.2              1    43     1         43
#> 12       2.0         0.5              2    43     1         43
#> 13       5.0         1.0              1    43     1         43
#> 14       5.0         0.2              1    43     1         43
#> 15       5.0         0.5              2    43     1         43
#> 16      10.0         1.0              1    43     1         43
#> 17      10.0         0.2              1    43     1         43
#> 18      10.0         0.5              2    43     1         43