def funcIDNormPCL(target, source, env, iMaxLines=100000):
strT, astrSs = sfle.ts(target, source)
strS = astrSs[1] if (
len(sfle.readcomment(astrSs[1])) > c_iSkip) else astrSs[0]
iLC = sfle.lc(strS)
return (sfle.ex("Normalizer -t pcl -T medmult < " + strS, strT) if
(iLC < iMaxLines) else sfle.ex("head -n 3 < " + strS, strT))
def funcMergeMap(target, source, env):
strT, astrSs = sfle.ts(target, source)
fileTaxa, fileMerge, fileIDRaw = astrSs[:3]
astrTaxa = sfle.readcomment(fileTaxa)
strMap = arepa.get_mappingfile(astrTaxa[0]) if astrTaxa else ""
return (sfle.ex([fileMerge, fileIDRaw, strMap, strT])
if strMap else sfle.ex(["cp", fileIDRaw, strT]))
def funcRawMap(target, source, env):
strT, astrSs = sfle.ts(target, source)
strGPLTXTGZ, strPlatformTXT, strProgAnnot2Map, strProgGPL2TXT = astrSs[:4]
strGPLID = (sfle.readcomment(open(strPlatformTXT)) or [""])[0]
return (ex([strProgAnnot2Map, strGPLTXTGZ, strT])
if not (sfle.isempty(str(strGPLTXTGZ))) else ex(
[strProgGPL2TXT, c_strGPLPath + strGPLID, strT]))
def funcRawProcess(target, source, env):
strT, astrSs = sfle.ts(target, source)
strIn, strRData, strExpMetadata, strCondMetadata = astrSs[:4]
iRet = sfle.ex((sfle.cat(strIn), "| R --no-save", "--args", strRData, strT,
c_strPPfun, strExpMetadata, strCondMetadata))
if iRet:
iRet = sfle.ex((sfle.cat(strIn), " | R --no-save --args", c_strFilePCL,
strT, c_strPPfun, strExpMetadata, strCondMetadata))
return iRet
def funcGPLsTXT( target, source, env ):
strT, astrSs = sfle.ts( target, source )
setstrGPLs = set()
for strLine in gzip.open( astrSs[0] ):
mtch = re.search( r'^!dataset_platform\s*=\s*(\S+)', strLine )
if mtch:
setstrGPLs.add( mtch.group( 1 ) )
with open( strT, "w" ) as fileOut:
fileOut.write( "%s\n" % "\n".join( ("-".join( (c_strID, s) ) for s in setstrGPLs) ) )
return None
def _makeDescription( target, source, env ):
strT, astrSs = sfle.ts( target, source )
pHash = pickle.load(open(astrSs[0]))
pHashDescription = { "Package": strDatasetName.replace("-", "."), "Type": "Package", "Title": pHash.get("title"),
"Version": arepa.c_strVersion, "Author": ", ".join(arepa.c_astrAuthors),
"Date": arepa.c_strDate, "Maintainer": arepa.c_strMaintainer,
"Depends": "R (>= 2.10.0), affy", "Suggests": "survival", "URL": arepa.c_strURL,
"License": arepa.c_strLicense, "Description": "ARepA generated package" }
with open(strT, "w") as outputf:
for k, v in list(pHashDescription.items()):
outputf.write( k + ": " + v + "\n" )
def funcIDsTXT(target, source, env):
strT, astrSs = sfle.ts(target, source)
astrFiles = sfle.ftpls(c_strHost, sfle.d(c_strPath, c_strID))
astrFiles = [
mtch.group(1) for mtch in [
_f for _f in (re.search(r'(GSE\d+(?:-GPL\d+)?)', s)
for s in astrFiles) if _f
]
]
with open(strT, "w") as fileOut:
fileOut.write("%s\n" % "\n".join(astrFiles))
return None
def _makeMasterMan( target, source, env ):
strT, astrSs = sfle.ts( target, source )
pHash = pickle.load(open(astrSs[0]))
def _metaStr( strDescription, strContent ):
return "\\"+ strDescription + "{" + strContent + "}"
strDataAccession = arepa.cwd( ) + "-package"
strDataTitle = pHash.get( "title" ) or ""
strDataGloss = pHash.get( "gloss" ) or ""
aastrOut = [("name", strDataAccession), ("title", strDataTitle), ("description", strDataGloss)]
with open( strT, "w" ) as outputf:
for strDescription, strContent in aastrOut:
outputf.write( _metaStr( strDescription, strContent ) + "\n" )
def funcRet(target,
source,
env,
setstrInclude=setstrInclude,
setstrExclude=setstrExclude):
strT, astrSs = sfle.ts(target, source)
for strS in astrSs:
for astrLine in csv.reader(open(strS), csv.excel_tab):
if not (astrLine and astrLine[0]):
continue
strID = astrLine[0]
if (setstrInclude and
(strID not in setstrInclude)) or (strID in setstrExclude):
continue
env["sconscript_child"](target, source[0], env,
c_strID + "-RAW")
def _compileR( target, source, env ): strT, astrSs = sfle.ts( target, source ) sfle.ex( ["chmod", "755", strDirectory] ) sfle.ex( ["R", "CMD", "build", strDirectory] ) with open( strT, "w" ) as outputf: outputf.write( "R package compiled OK")
def funcIDQUANT(target, source, env):
strT, astrSs = sfle.ts(target, source)
strS = astrSs[0]
iLC = sfle.lc(strS)
return (sfle.ex("echo '-1.5\t-0.5\t0.5\t1.5\t2.5\t3.5\t4.5' >" + strT))
def funcIDDAB(target, source, env):
strT, astrSs = sfle.ts(target, source)
strS = astrSs[0]
iLC = sfle.lc(strS)
return (sfle.ex((sfle.cat(strS), " | Distancer -o", strT)) if
(iLC > 3) else sfle.ex("echo", strT))
def funcIDKNNPCL(target, source, env, iMaxLines=40000):
strT, astrSs = sfle.ts(target, source)
strS = astrSs[0]
iLC = sfle.lc(strS)
return (sfle.ex("KNNImputer < " + strS, strT) if
(iLC < iMaxLines) else sfle.ex("head -n 3 < " + strS, strT))
def _funcDAB(target, source, env):
strT, astrSs = sfle.ts(target, source)
strOut, strMap = astrSs[:2]
return sfle.ex(("Dat2Dab", "-o", strT, "-i",
(strOut if sfle.isempty(strMap) else strMap)))
