alphabetr (version 0.2.2)

bagpipe: Identify candidate alpha/beta pairs.

Description

bagpipe() uses the alphabetr resampling procedure on sequencing data to identify candidate alpha/beta pairs. The procedure takes a subsample of the data without replacement, calculates association scores with chain_scores, and then for each well, uses the Hungarian algorithm to determine the most likely pairings for the chains found in the well. Each time this is done is a replicate, and the number of replicates is specified as an option. A threshold is then used to filter the candidate pairs that appear in proportion of the replicates larger than the threshold, resulting in the final list of candidate pairs. Bagpipe is an acronym for bootstrapping alphabetr generated pairs procedure (based on older versions that utilized bootstrapping)

Usage

bagpipe(alpha, beta, replicates = 100, frac = 0.75, bootstrap = FALSE)

Arguments

alpha

Matrix recording which alpha chains appear in each well of the data. See create_data.

beta

Matrix recording which beta chains appear in the each well of the data. See create_data.

replicates

The number of times the resampling procedure is repeated, i.e. the number of replicates. At least 100 replicates is recommended.

frac

The fraction of the wells resampled in each replicate. Default is 75% of the wells

bootstrap

Legacy option. Calls a bootstrapping strategy (which resamples with replacement) instead of sampling a subset without replacement.

Value

A n x 5 matrix where n is the number of clones determined by bagpipe(). Each row represents the chains of the clone. Columns 1 and 2 represent the beta chain indices of the clone. Columns 3 and 4 represent the alpha chain indices of the clone. Column 5 represents the number of replicates that this clone was found in. Column 5 is used to filter our the clones that have not be determined by a certain "threshold" proportion of replicates.Note that column 1 == column 2 and column 3 == column 4 since bagpipe() does not try to determine dual-alpha or dual-beta clones.

Examples

 # see the help for create_clones() and create_data()
 clones <- create_clones(numb_beta = 1000,
                         dual_alpha = .3,
                         dual_beta = .06,
                         alpha_sharing = c(0.80, 0.15, 0.05),
                         beta_sharing  = c(0.75, 0.20, 0.05))
 dat <- create_data(clones$TCR, plate = 5,
                    error_drop = c(.15, .01),
                    error_seq  = c(.05, .001),
                    error_mode = c("lognormal", "lognormal"),
                    skewed = 10,
                    prop_top = 0.6,
                    dist = "linear",
                    numb_cells = matrix(c(50, 480), ncol = 2))

 ## Not run: 
#  # normally want to set replicates to 100 or more
#  pairs <- bagpipe(alpha = dat$alpha,
#                beta  = dat$beta,
#                replicates = 5,
#                frac = 0.75,
#                bootstrap = FALSE)
# 
#  # using a threshold of 0.3 of replicates
#  pairs <- pairs[pairs[, 5] > 0.3, ]
#  ## End(Not run)