Add readAIRR() and update readChangeoDb(). They should accept a vector of files and check uniqueness

[ssnn]

Update alakazam::readChangeoDb's file argument to accept a vector and behave accordingly if length(file) > 1.

For populating sample_id, require file to be a named vector and use the names for the sample_id, with some argument to control whether this action is performed. Eg:

• samples = NULL: leave `sample_id` alone
• samples = "auto": populate `sample_id` from names(file). Use numbers if is.null(names(file)).
• samples = character vector where length(samples) == length(files): use these as sample_id values for each file.

uniquify=TRUE could either append a number or the sample_id to the relevant columns.  You'll probably want to also check for, and fix, uniqueness within a single file (could just be a numeric suffix in that case).

Make a alakazam::readAIRR, which wraps airr::read_rearrangement and does all the same stuff.

You should be able to abstract the AIRR/Changeo functions too. As the only difference seems to be column names and readr::read_tsv vs airr::read_rearrangement.

[Jason Vander Heiden]

This is currently what I do, in case it helps:

# Imports
library(tidyverse)
library(alakazam)
library(airr)

# Find AIRR files
files <- dir(DATA_PATH, pattern="db-pass.tsv", full.names=T, recursive=T)
names(files) <- str_extract(files, "SAM\\d+")

# Load AIRR files
db <- lapply(files, read_rearrangement) %>%
    bind_rows(.id="sample_id") %>%
    mutate(sequence_id=paste(sequence_id, sample_id, sep="_"),
           cell_id=paste(cell_id, sample_id, sep="_"),
           chain=getChain(locus))

# Filter to productive and remove doublets
db <- db %>%
    filter(productive, !is.na(chain)) %>%
    group_by(sample_id, cell_id, chain) %>%
    dplyr::slice(which.max(umi_count))

‌