Diploid Copy Number Estimation

Computes per-sample diploid copy number by normalizing each sample’s read counts relative to its nearest neighbors, accounting for sample-specific sequencing depth.

grid.utils.compute_dipcn.compute_diploid_genotypes(config, console)

Compute diploid copy number estimates.

Two modes depending on config:

Mode 1 — GC normalize enabled (gc_normalize: run: True):

Writes corrected_rpk_ratio directly from the GC normalization output to the diploid genotypes file. No neighbor normalization applied. This gives the best correlation with Bionano optical mapping (r=0.986).

Mode 2 — GC normalize disabled (gc_normalize: run: False):

Replicates Mukamel’s awk normalization exactly:

norm_reads = reads[sample] / sample_scale
             / mean( reads[neighbor] / neighbor_scale )

where the mean is over the top n_nbr neighbors.

Return type:

None

grid.utils.compute_dipcn.load_neighbors(neighbors_file)

Parse the gzipped neighbors file.

Each line format:

sample_id sample_scale nbr1_id nbr1_scale nbr1_norm_dist nbr2_id …

Returns:

{sample_id: [(neighbor_id, neighbor_scale), …]} sample_scales : {sample_id: sample_scale}

Return type:

neighbors

Parameters:

neighbors_file (Path)