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" )
import arepa import gzip import os import re import sfle import sys def test( iLevel, strID, hashArgs ): return ( iLevel == 1 ) and ( strID.find( "GDS" ) == 0 ) if "testing" in locals( ): sys.exit( ) pE = DefaultEnvironment( ) c_strID = arepa.cwd( ) c_fileIDTXT = sfle.d( pE, c_strID + ".txt" ) c_fileIDSOFTGZ = sfle.d( pE, c_strID + ".soft.gz" ) Import( "hashArgs" ) #=============================================================================== # Download SOFT file #=============================================================================== sfle.download( pE, hashArgs["c_strURLGDS"] + os.path.basename( str(c_fileIDSOFTGZ) ) ) def funcGPLsTXT( target, source, env ): strT, astrSs = sfle.ts( target, source ) setstrGPLs = set() for strLine in gzip.open( astrSs[0] ):
def funcGeneIDMapping(pE,
fileDATin,
strGeneFrom=None,
fileLOGout=None,
strMAPin=None,
aiCOL=[0, 1],
iSkip=0,
iLevel=2):
try:
int(strMAPin)
strTaxID, strMAPin = strMAPin, None
except (ValueError, TypeError):
strTaxID = None
# Try to find taxid
if not strTaxID:
astrMatch = re.findall(r'taxid_([0-9]+)', c_strID)
strTaxID = astrMatch[0].strip() if astrMatch else None
if not strGeneFrom:
#if strGeneFrom is not specified, try to retrieve it from
#the "manual_geneids.txt" file in the respective etc directory, else None
#this problem tends to arise in mixed microbial network data
if strTaxID and m_hashGeneIDs:
strOrg = " ".join(arepa.taxid2org(strTaxID).split(" ")[:iLevel])
strGeneFrom = m_hashGeneIDs.get(strOrg)
if not (strMAPin):
strDirAuto = c_strDirData if (c_strID == c_strPathRepo) else ""
strAutoMAPtmp = sfle.d(strDirAuto, c_strID + c_strSufMap)
strAutoMAP = strAutoMAPtmp if os.path.exists(strAutoMAPtmp) else None
# Use manual mapping files, files that are deeper down have priority
#else try automatically generated mapping file
afileMaps = Glob(sfle.d(c_strPathTopMapping, "*" + c_strSufMap))
strTopMAP = str(afileMaps[0]) if afileMaps else None
strMAPManTmp = sfle.d(c_strDirManMap, c_strID + c_strSufMap)
strMAPin = (strMAPManTmp if os.path.exists(strMAPManTmp) else
None) or strTopMAP or strAutoMAP
# Use provided mapping files
if not (strMAPin) and strTaxID:
for fileMAPname in Glob(
sfle.d(c_strPathUniprotKO, "*" + c_strSufMap)):
if re.search(r'\D' + strTaxID + r'\D', str(fileMAPname)):
strMAPin = str(fileMAPname)
break
# Else ask arepa to figure out an appropriate mapping file
if not (strMAPin) and strTaxID:
strMAPin = arepa.get_mappingfile(strTaxID)
# Else use BridgeDB files
if not (strMAPin) and strTaxID:
for strSpeciesName in list(c_pHashBridgeDB.keys()):
if strSpeciesName in arepa.taxid2org(strTaxID):
strMAPin = c_pHashBridgeDB[strSpeciesName]
break
strBase, strExt = os.path.splitext(str(fileDATin))
strCount = funcCounter(g_iterCounter)
strT = strBase + c_strMapped + strCount + strExt
aastrPrefix = [[str(fileDATin)], [True, strT], "-c", str(aiCOL)]
astrGeneFrom = ["-f", strGeneFrom] if strGeneFrom else ["-x"]
astrGeneTo = ["-t", c_astrGeneTo[0]]
astrSkip = ["-s", iSkip]
afileLOGout = ["-l", [True, str(fileLOGout)]] if fileLOGout else []
astrMapIn = ["-m", [strMAPin]] if strMAPin else []
aastrArgs = aastrPrefix + astrGeneFrom + astrGeneTo + astrSkip + afileLOGout + astrMapIn
pE.Precious(strMAPin)
return pE.Depends(
sfle.op(pE, c_funcGeneMapper, aastrArgs),
sfle.scons_child(
pE, c_strPathGeneMapper, None, None,
None, ([strMAPin] if not (sfle.in_directory(
strMAPin, os.path.abspath(arepa.cwd()))) else None)))
substantial portions of the Software.
THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR IMPLIED,
INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.
IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY,
WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE
OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
"""
import arepa
import csv
import re
import sfle
import sys
c_repo = arepa.cwd()
c_strTaxid = c_repo + "_taxid_"
c_strMode = "mode_"
def symbol(hashSymbols, strValue):
return hashSymbols.setdefault(strValue, len(hashSymbols))
if len(sys.argv) < 1:
raise Exception("Usage: string2c.py [taxa] < <string.txt>")
iMin = int(sys.argv[1])
strTaxa = None if (len(sys.argv) <= 2) else sys.argv[2]
setTaxa = arepa.taxa(strTaxa)
import sfle
import sys
import ftplib
import re
def test(iLevel, strID, hashArgs):
return (iLevel == 3) and (strID.find("GSE") == 0)
if "testing" in locals():
sys.exit()
pE = DefaultEnvironment()
c_strID = arepa.cwd().replace("-RAW", "")
c_strURLGEO = 'ftp.ncbi.nih.gov'
c_strURLGEOsupp = 'pub/geo/DATA/supplementary/samples/'
c_strURLSupp = 'ftp://' + c_strURLGEO + '/' + c_strURLGEOsupp
c_strFileGSM = "../GSM.txt"
c_strFilePCL = "../" + c_strID + ".pcl"
c_listTs = sfle.readcomment(c_strFileGSM)
c_fileProgReadCel = sfle.d(pE, arepa.path_repo(), sfle.c_strDirSrc,
"readCel.R")
c_fileProgProcessRaw = sfle.d(pE, arepa.path_repo(), sfle.c_strDirSrc,
"preprocessRaw.R")
c_strInputRData = arepa.cwd() + ".RData"
c_strOutputRData = c_strInputRData.replace("-RAW", "")
c_filePPfun = sfle.d(pE, arepa.path_repo(), sfle.c_strDirEtc, "preprocess")
IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY,
WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE
OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
"""
import arepa
import os
import sfle
import sys
c_afileSConscripts = sorted(
Glob(sfle.d(arepa.path_repo(), sfle.c_strDirSrc, "SConscript*")))
if os.path.isfile("SConscript"):
exec(compile(open("SConscript").read(), "SConscript", 'exec'))
else:
hashArgs = {}
# I tried very hard to do this using import, but I can't find a way to prematurely
# halt an import without sys.exit, which kills the entire process.
for fileSConscript in c_afileSConscripts:
strSConscript = fileSConscript.get_abspath()
hashEnv = {"test": lambda *a: False, "testing": True}
try:
exec(compile(open(strSConscript).read(), strSConscript, 'exec'),
hashEnv)
except SystemExit:
pass
if hashEnv["test"](arepa.level(), arepa.cwd(), hashArgs):
exec(compile(open(strSConscript).read(), strSConscript, 'exec'))