Remove excess pairs for individual single chains

This function filters the paired TCR data, keeping at most one beta-chain partner for each individual alpha-chain and at most two alpha-chain partners for each individual beta-chain.

Usage

clean_pairs(
  data,
  n_max_alpha = 2,
  n_max_beta = 1,
  prefer_functional = TRUE,
  verbose = TRUE
)

Arguments

data

a TIRTLseqDataSet object

n_max_alpha

(optional) the maximum number of alpha chains allowed paired with an individual beta chain (default 2)

n_max_beta

(optional) the maximum number of beta chains allowed paired with an individual alpha chain (default 1)

prefer_functional

(optional) if TRUE, prefer functional to non-functional chains even if they are less frequent

verbose

(optional) whether to print progress of the function (default is TRUE).

Value

Returns a TIRTLseqDataSet object with excess pairs removed.

Details

Excess partners for an individual chain are often sequencing errors of the true partner. The sequencing error-derived chains often share the same V/J segment as the true partner, but are found at much lower read fractions.

As a heuristic, to mitigate this phenomenon, for each unique beta chain we group the partner alpha chains by their V/J segments and keep only the alpha chains with the highest read fraction for each group. Out of the remaining alpha chain partners, we keep up to two chains (can be changed with n_max_alpha) with the highest read fractions.

We go through a similar process with each unique alpha chain, grouping partner beta chains by their V/J segments. However, we keep only one beta chain (can be changed with n_max_beta).

See also

Other data_processing: TIRTL_process(), add_single_chain_data(), filter_duplicate_tcrs(), filter_mait(), filter_nonfunctional_TCRs(), filter_short_cdr3s(), filter_v_alleles(), identify_non_functional_seqs(), identify_paired(), prep_for_tcrdist(), remove_duplicates()