diffHic: Differential Analysis of Hi-C Data

# Tests for zero-inputs into various diffHic functions.

suppressWarnings(suppressPackageStartupMessages(require(diffHic)))

# Testing what happens with an empty input file.

f.out <- "empty.h5"
param <- pairParam(GRanges("chrA", IRanges(1:5, 1:5)))
savePairs(data.frame(anchor1.id=integer(0), anchor2.id=integer(0)), file=f.out, param=param)

loadChromos(f.out) # While we're here, let's see what happens.

squareCounts(f.out, param)

marginCounts(f.out, param)

totalCounts(f.out, param)

connectCounts(f.out, param, GRanges("chrA", IRanges(1, 4)))

extractPatch(f.out, param, GRanges("chrA", IRanges(1, 4)), width=10)

unlink(f.out)

# Testing with an empty InteractionSet.

ghost <- InteractionSet(list(counts=matrix(0, nrow=0, ncol=1)), 
    GInteractions(integer(0), integer(0), regions=GRanges("chrA", IRanges(1:5, 1:5)), mode="reverse"),
    colData=DataFrame(totals=1e6))

getArea(ghost)
getArea(ghost, bp=TRUE)

filterDirect(ghost)
filterDirect(ghost, reference=ghost)

filterTrended(ghost)
try(filterTrended(ghost, reference=ghost)) # This will fail, as interpolation is impossible.

filterPeaks(ghost, integer(0))

enrichedPairs(ghost)

try(compartmentalize(ghost)) # This will fail, as interpolation is impossible.
try(compartmentalize(ghost, dist.correct=FALSE)) # This will also fail, due to non-unique k-means.

correctedContact(ghost)

ghost.ranges <- SummarizedExperiment(matrix(0, 0, 1), GRanges(), colData=DataFrame(totals=1e6))
try(normalizeCNV(ghost, ghost.ranges)) # locfit isn't as robust as loessFit
ghost.ranges$totals <- NULL
try(normalizeCNV(ghost, ghost.ranges)) # spits the dummy when totals are not the same.

matchMargins(ghost, ghost.ranges)

asDGEList(ghost)

diClusters(ghost, data.frame(PValue=integer(0), logFC=numeric(0)), target=0.05, cluster.args=list(tol=1))

annotatePairs(ghost, indices=integer(0), regions=GRanges())