CAMDAC Deconvolution Only

Deconvolution only

The CAMDAC equation can be used to infer pure tumour DNA methylation rates, provided the following information is available per CpG:

• Bulk tumour methylation rates
• Tumour copy number state
• Tumour purity

Here is an example for 5 CpGs from a single sample. Note: the normal copy number state is assumed diploid (2) in humans:

# Set parameters
bulk = c(0.3, 0.5, 0.2, 0.1, 0.9)
normal = c(0.3, 0.9, 0.1, 0.7, 0.5)
ploidy = c(2, 2, 1, 3, 4)
purity = 0.8

# Deconvolve methylation rates
pure_meth = CAMDAC:::calculate_mt(bulk, normal, purity, ploidy)

# Set clean rates based on threshold
pure_meth_clean = dplyr::case_when(
  pure_meth < 0 ~ 0,
  pure_meth > 1 ~ 1,
  TRUE ~ pure_meth
)

After deconvolution, it may be useful to estimate the CpG coverage in the deconvolved tumour sample. Additionally, the highest density interval (HDI) of the methylation rate may be informative for quality control. These metrics can be calculated given additional information on bulk methylated and unmethylated read counts:

# Optional: calculate effective coverage of the tumour
# # Requires coverage per CpG in the bulk sample
bulk_coverage = c(10, 20, 5, 15, 30)
pure_effective_coverage = CAMDAC:::calculate_mt_cov(bulk_coverage, purity, ploidy)

# Optional: calculate the HDI of the pure tumour methylation rate
bulk_methylated_count = c(3, 10, 1, 2, 27)
bulk_unmethylated_count = c(7, 10, 4, 13, 3)
normal_methylated_count = c(3, 9, 1, 5, 2)
normal_unmethylated_count = c(7, 11, 3, 8, 3)

# HDI function (fast)
CAMDAC:::hdi_norm_approx(
  bulk_methylated_count,
  bulk_unmethylated_count,
  normal_methylated_count,
  normal_unmethylated_count,
  purity,
  ploidy
)

# HDI function (most accurate)
CAMDAC:::vec_HDIofMCMC_mt( 
  bulk_methylated_count,
  bulk_unmethylated_count,
  normal_methylated_count,
  normal_unmethylated_count,
  purity,
  ploidy,
  credMass=0.99
)