# test_GeneHancerDB.R
#------------------------------------------------------------------------------------------------------------------------
(ghdb)
(RUnit)
#------------------------------------------------------------------------------------------------------------------------
(!("ghdb")){
ghdb <- ()
} # creating for several tests below
#------------------------------------------------------------------------------------------------------------------------
runTests <- ()
{
test_ctor()
test_listTissues()
test_.eliminateDupsCollapseTissues()
test_foxo6()
test_klf1()
test_failureByTissue()
test_queryUnknownGene()
test_getEnhancers()
test_queryByRegion()
#test_to.hg19()
} # runTests
#------------------------------------------------------------------------------------------------------------------------
test_ctor <- ()
{
(("--- test_ctor"))
checkTrue("GeneHancerDB" %in% (ghdb))
} # test_ctor
#------------------------------------------------------------------------------------------------------------------------
test_listTissues <- ()
{
(("--- test_listTissues"))
tissues <- (ghdb, "MEF2C")
checkTrue((tissues) > 100)
checkTrue((("placenta", "Placenta") %in% tissues))
tissues.gata2 <- (ghdb, targetGene="GATA2")
# 120-130 with gh 411, 152 with v50
checkTrue((tissues.gata2) > 150)
checkTrue((tissues.gata2) < 160)
} # test_listTissues
#------------------------------------------------------------------------------------------------------------------------
test_.eliminateDupsCollapseTissues <- ()
{
(("--- test_.eliminateDupsCollapseTissues"))
((package="TrenaProjectHG38", "extdata", "testing", "tbl.gh.results.toTestReductionPreservingTissues.RData"))
checkEquals((tbl), (956, 17))
tbl.trimmed <- ghdb:::.eliminateDupsCollapseTissues(tbl)
checkEquals((tbl.trimmed), (50, 16))
# do a rough assay of the collapsed tissue column
spread <- ((tbl.trimmed$tissue)) # [1] 16 62 156 338 2213
checkTrue(spread[1] < 20)
checkTrue(spread[5] > 2000)
} # test_.eliminateDupsCollapseTissues
#------------------------------------------------------------------------------------------------------------------------
# a simple test case, chosen for historical (and no longer important) reasons
test_foxo6 <- ()
{
(("--- test_foxo6"))
# ENCODE and Ensembl use different capitalization
tissues <- ("placenta", "Placenta")
tbl <- (ghdb, "FOXO6", tissues)
checkTrue((tbl))
checkTrue((tbl) > 10) # 14 for 5.16, 19 for v54, 23 for v50, 13 for v411
checkTrue((tissues %in% tbl$tissue))
checkEquals((((tbl$sig))), 0) # no duplicated regions
checkTrue(((";", tbl$tissue)) > 0) # 4 instances of "Placenta;placenta"
tbl.all <- (ghdb, "FOXO6", tissues="all")
checkTrue((tbl.all))
checkTrue((tbl.all) > 40) # 44 for v5.16 3 for v411, 71 for v50, 70 for v54
checkEquals((((tbl.all$sig))), 0) # no duplicated regions
checkTrue(((tbl.all$tissue)) > 1200) # 2151 for v54
checkTrue(((";", tbl.all$tissue)) > 40) # most tissue values are semicolo separated multiples
} # test_foxo6
#------------------------------------------------------------------------------------------------------------------------
# another sample, with KLF1 important in erythropoiesis
test_klf1 <- ()
{
(("--- test_klf1"))
# ENCODE and Ensembl use different capitalization
allTissues <- (ghdb, "KLF1")
tissues <- (("myeloid", allTissues, ignore.case=, v=),
("eryth", allTissues, ignore.case=, v=),
("blood", allTissues, ignore.case=, v=))
tbl <- (ghdb, "KLF1", tissues)
(tbl)
checkTrue((tbl) >= 3)
} # test_klf1
#------------------------------------------------------------------------------------------------------------------------
# a simple test case, chosen for historical (and no longer important) reasons
test_failureByTissue <- ()
{
(("--- test_failureByTissue"))
trem2.tissues <- (ghdb, "TREM2", tissues="all")$tissue
tissues <- ("brain", "Brain")
checkTrue(((tissues, trem2.tissues)) == 0)
(
tbl <- (ghdb, "TREM2", tissues)
)
checkEquals((tbl), 0)
# double check now that at least one tissue reports for TREM2
tbl <- (ghdb, "TREM2", "placenta")
checkTrue((tbl) > 0)
} # test_failureByTissue
#------------------------------------------------------------------------------------------------------------------------
test_queryUnknownGene <- ()
{
(("--- test_queryUnknownGene"))
<- "LINC00982"
(
tbl <- (ghdb, , tissues="all")
)
checkEquals((tbl), (0,0))
} # test_queryUnknownGene
#------------------------------------------------------------------------------------------------------------------------
test_getEnhancers <- ()
{
(("--- test_getEnhancers"))
all.tissues <- (ghdb, "TREM2")
tbl.trem2.all <- (ghdb, "TREM2")
checkTrue((tbl.trem2.all$geneSymbol == "TREM2"))
(tbl.trem2.all)
checkTrue((tbl.trem2.all) > 8) # gh50: 16 gh54: 16
tbl.mef2c <- (ghdb, "MEF2C")
checkTrue((tbl.mef2c$geneSymbol == "MEF2C"))
checkTrue((tbl.mef2c) > 10) # 5.16: 13, gh411: 12 , gh50: 201, gh54 206
(tbl.bogus <- (ghdb, "bogus99"))
checkEquals((tbl.bogus), 0)
#----------------------------------------------------------------------------------
# hoxa5 has a nonsensical promoter, 86kb in size
# simon fishilevich calls this an outlier, worth elminating
# comes from ENCODE. he suggests this filter:
# drop all elements >=10kb AND (overlap a gene TSS OR overlap >=3 gene exons)
# i am implementing just the first clause for now
#----------------------------------------------------------------------------------
tbl.hoxa5 <- (ghdb, "HOXA5", maxSize=100000)
checkTrue((tbl.hoxa5) > 10)
sizes <- (tbl.hoxa5, 1 + - )
checkEquals(((sizes > 10000)), 2) # gh411: 1 gh50: 2, gh54: 2
tbl.hoxa5 <- (ghdb, "HOXA5", maxSize=10000)
checkTrue((tbl.hoxa5) > 8) # 9 for gh5.16, 10 with gh411
sizes <- (tbl.hoxa5, 1 + - )
checkEquals(((sizes > 10000)), 0)
} # test_getEnhancers
#------------------------------------------------------------------------------------------------------------------------
test_queryByRegion <- ()
{
(("--- test_queryByRegion"))
chrom <- 'chr1'
<- 173595097
<- 173692966
tbl <- (ghdb, chrom, , )
checkTrue((tbl) > 50) # gh54: 76 gh59: 80
checkTrue((tbl$chrom == chrom))
checkTrue((tbl$ >= ))
checkTrue((tbl$ <= ))
tbl.best <- (tbl, combinedscore > 250)
checkTrue((tbl.best) > 10) # 19 in gh5.16, 15 with gh version 5.0 > 500, 19 >250 in gh54
checkTrue((tbl.best) < 30)
} # test_queryByRegion
#------------------------------------------------------------------------------------------------------------------------
test_to.hg19 <- ()
{
(("--- test_to.hg19"))
tbl.all <- (ghdb, "FOXO6", tissues="all")
checkEquals((tbl.all), 70) # gh50=71, gh54=70
tbl.hg19 <- (ghdb, tbl.all)
checkEquals((tbl.all), (tbl.hg19))
checkTrue(((tbl.all$, tbl.hg19$)) == 0)
checkTrue((tbl.all$chrom == tbl.hg19$chrom))
} # test_to.hg19
#------------------------------------------------------------------------------------------------------------------------
test_brca1 <- ()
{
(("--- test_brca1"))
tbl.brca1.tissue <- (ghdb, "BRCA1", "mammary epithelial cell")
checkTrue((tbl.brca1.tissue) < 5)
tbl.brca1.all <- (ghdb, "BRCA1", "all")
checkTrue((tbl.brca1.all) > 15)
} # test_to.hg19
#------------------------------------------------------------------------------------------------------------------------
(!())
runTests()
