tests/testthat/testing-data.R

In BimberLab/cellhashR: A Package for Demultiplexing Cell Hashing Data

tests <- list(
	'282-1' = list(
		input = '../testdata/cellHashing/282-1-HTO_cellHashingRawCounts.txt',
		htos = paste0('HTO-', c(2:3, 8, 10, 12)),
		gexBarcodeFile = '../testdata/cellHashing/282-1-whitelist.txt',
		CalledCells = 3476,
		Singlet = 2198,
		Doublet = 947,
		MultiSeqCalled = 4010,
		Discordant = 1524,
		SeuratCalled = 3179,
		minAllowableDoubletRateFilter = 0.4
	),
	'283' = list(
		input = '../testdata/cellHashing/283-cellbarcodeToHTO.calls.citeSeqCounts.txt',
		htos = paste0('HTO-', c(2:6)),
		gexBarcodeFile = '../testdata/cellHashing/283-validCellIndexes.csv',
		CalledCells = 3600,
		Singlet = 2247,
		Doublet = 723,
		MultiSeqCalled = 3223,
		Discordant = 1400,
		SeuratCalled = 4116,
		minAllowableDoubletRateFilter = 0.4
	),
	'438-21' = list(
		input = '../testdata/438-21-GEX/umi_count',
		citeSeqCountDir = '../testdata/438-21-GEX/',
		htos = paste0('MS-', c(11, 12)),
		CalledCells = 4833,
		Singlet = 3971,
		Doublet = 366,
		MultiSeqCalled = 4504,
		Discordant = 167,
		SeuratCalled = 4096
	),
	'438-24' = list(
		input = '../testdata/438-24-GEX/umi_count/',
		htos = paste0('MS-', c(11, 12)),
		CalledCells = 4624,
		Singlet = 3920,
		Doublet = 251,
		MultiSeqCalled = 4547,
		Discordant = 376,
		SeuratCalled = 3038,
		BffQuantile = 4941
	),
	'449-1' = list(
		input = '../testdata/449-1-GEX/umi_count',
		htos = paste0('MS-', c(2:16)),
		CalledCells = 1106,
		Singlet = 986,
		Doublet = 95,
		MultiSeqCalled = 1081,
		Discordant = 0,
		SeuratCalled = 1103
	),
	'457-1' = list(
		input = '../testdata/457-1-GEX/umi_count/',
		htos = paste0('MS-', c(1:3, 5:8)),
		CalledCells = 2437,
		Singlet = 2060,
		Doublet = 323,
		MultiSeqCalled = 2383,
		Discordant = 0,
		SeuratCalled = 2432
	),
	'471-1' = list(
		input = '../testdata/471-1-GEX/umi_count',
		htos = paste0('MS-', c(1, 2)),
		CalledCells = 5000,
		Singlet = 3368,
		Doublet = 580,
		MultiSeqCalled = 5000,
		Discordant = 0,
		SeuratCalled = 3948
	),
	'471-2' = list(
		input = '../testdata/471-2-GEX/umi_count',
		htos = paste0('MS-', c(3, 4)),
		CalledCells = 5000,
		Singlet = 2085,
		Doublet = 587,
		MultiSeqCalled = 38,
		Discordant = 0,
		SeuratCalled = 2672
	),
	'483-3' = list(
		input = '../testdata/483-3-GEX/umi_count',
		htos = paste0('MS-', c(2:4, 6:8, 10:13)),
		CalledCells = 166,
		Singlet = 148,
		Doublet = 8,
		MultiSeqCalled = 156,
		Discordant = 0,
		SeuratCalled = 165
	)
)

DoTest <- function(test, methods = c('multiseq', 'htodemux'), skipNormalizationQc = FALSE) {
	barcodeFile <- test$input
	barcodeData <- ProcessCountMatrix(rawCountData = barcodeFile, barcodeWhitelist = test$htos)

	if (nrow(barcodeData) == 0) {
		stop('No passing HTOs')
	}

	if (ncol(barcodeData) == 0) {
		stop('No passing cells')
	}

	#Subset to keep reasonable
	if (ncol(barcodeData) > 5000) {
		print('Subsetting barcodeData to 5000 cells')
		barcodeData <- barcodeData[,1:5000]
	}

	# This is giving memory errors on github runners:
	if (!skipNormalizationQc) {
		PlotNormalizationQC(barcodeData)
	}

	metricsFile <- 'metrics.txt'
	if (file.exists(metricsFile)) {
		unlink(metricsFile)
	}

	minAllowableDoubletRateFilter <- 0.15
	if ('minAllowableDoubletRateFilter' %in% names(test)) {
		minAllowableDoubletRateFilter <- test[['minAllowableDoubletRateFilter']]
	}

	# NOTE: extraneous argument added, since at one point this broke caller-specific argument parsing
	df <- GenerateCellHashingCalls(barcodeMatrix = barcodeData, methods = methods, metricsFile = metricsFile, extraArgument = 'foo', minAllowableDoubletRateFilter = minAllowableDoubletRateFilter)

	# Ensure actual consensus callers reported correctly:
	expect_true('MethodForConsensus' %in% names(df))
	expect_equal(1, length(unique(df$MethodForConsensus)))

	return(list(barcodeData = barcodeData, df = df, metricsFile = metricsFile))
}