示例#1
0
import gzip


def test(iLevel, strID, hashArgs):
    return (iLevel == 2) and (strID.find("GDS") == 0)


if "testing" in locals():
    sys.exit()

pE = DefaultEnvironment()

c_strID = arepa.cwd()
c_strGDS, c_strGPL = c_strID.split("-")[:2]

c_fileInputSConscript = sfle.d(pE, arepa.path_arepa(), sfle.c_strDirSrc,
                               "SConscript_pcl-dab.py")
c_fileRSConscript = sfle.d(pE, arepa.path_arepa(), sfle.c_strDirSrc,
                           "SConscript_rpackage.py")
c_fileInputSOFTGZ = sfle.d(pE, "../" + c_strGDS + ".soft.gz")
c_fileInputManCurTXT = sfle.d(pE, arepa.path_repo(), sfle.c_strDirEtc,
                              "manual_curation/", c_strID + ".txt")
c_filePPfun = sfle.d(pE, arepa.path_repo(), sfle.c_strDirEtc, "preprocess")
c_strPPfun = sfle.readcomment(c_filePPfun)[0]

c_fileTaxa = sfle.d(pE, "taxa.txt")
c_fileStatus = sfle.d(pE, "status.txt")
c_filePlatform = sfle.d(pE, "platform.txt")
c_fileIDMap = sfle.d(pE, c_strID + ".map")
c_fileIDMapRaw = sfle.d(pE, c_strID + "_raw.map")
c_fileIDPKL = sfle.d(pE, c_strID + ".pkl")
示例#2
0
	Maintainer: $arepa_maintainer
	Description: Automatically generated R package by arepa 
	Depends: R (>= 2.10.0), affy
	Suggests: survival
	License: MIT
	URL: http://huttenhower.sph.harvard.edu/arepa
"""

import sfle 
import arepa
import sys  
import pickle

c_strNAMESPACE		= r"'exportPattern("'"^[[:alpha:]]+"'")'"

c_fileProgUnpickle	= sfle.d( pE, arepa.path_arepa(), sfle.c_strDirSrc, "unpickle.py" )

def funcCheckRStructure( pE, strDatasetName, filePKL, fileNAMESPACE, fileDESCRIPTION, fileManMaster, strNAMESPACE = c_strNAMESPACE ):
	'''
	Completes necessary components for R package building 
	Assumes that data/ and man/ directories have the corresponding data and manual files per dataset
	
	Input: 
	fileNAMESPACE = pointer to NAMESPACE file to be tracked 
	fileManMaster = pointer to the master manual file in man/ 
	'''

	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"), 
示例#3
0
        if strName.find("processed-data") >= 0:
            return True
    return False


if "testing" in locals():
    sys.exit()

#Import( "pE" )
#Import( "hashArgs" )
c_strID = arepa.cwd()
c_strPrefix, c_strSuffix = c_strID.split("_")
c_strType = c_strID[2:6]
c_strInputIDSDRF = hashArgs["strFileIDSDRF"]
c_astrInputADFs = hashArgs["astrFileADFs"]
c_strInputSConscript = arepa.d(arepa.path_arepa(), arepa.c_strDirSrc,
                               "SConscript_pcl-dab.py")
c_strFileIDRawPCL = c_strID + "_00raw.pcl"
c_strFileIDNormPCL = c_strID + "_01norm.pcl"
c_strProgSamples2PCL = arepa.d(arepa.path_repo(), arepa.c_strDirSrc,
                               "samples2pcl.py")
c_fileInputData = Glob("../" + c_strPrefix + "*processed-data*" + c_strSuffix +
                       "*")[0]

#===============================================================================
# Calculate the final PCL + DAB
#===============================================================================

#- Map probe IDs and add PCL formatting
arepa.pipe(pE, c_fileInputData, c_strProgSamples2PCL, c_strFileIDRawPCL,
           [[True, s] for s in ([c_strInputIDSDRF] + c_astrInputADFs)])
示例#4
0
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.

SConscript_dat-dab.py: 

shared code across modules 
handles all behavior pertaining to 
mapping dat to dab and generating quant files 
"""

import sfle
import arepa

c_fileFlagSleipnir = sfle.d(pE, arepa.path_arepa(), sfle.c_strDirEtc,
                            "sleipnir")


def funcDAB(pE, fileOutDAB, afileInDAT):

    astrSleipnir = sfle.readcomment(c_fileFlagSleipnir)
    bSleipnir = (astrSleipnir[0] == "True")
    print("sleipnir", ("On" if bSleipnir else "Off"))

    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)))
示例#5
0
    for strName in pZIP.namelist():
        if strName.find("sample_table") >= 0:
            return True
    return False


if "testing" in locals():
    sys.exit()

#Import( "pE" )
#Import( "hashArgs" )
c_strID = arepa.cwd()
c_strType = c_strID[2:6]
c_strInputIDSDRF = hashArgs["strFileIDSDRF"]
c_astrInputADFs = hashArgs["astrFileADFs"]
c_strInputSConscript = arepa.d(arepa.path_arepa(), arepa.c_strDirSrc,
                               "SConscript_pcl-dab.py")
c_strFileIDRawTXT = c_strID + "_00raw.txt"
c_strFileIDRawPCL = c_strID + "_01raw.pcl"
c_strFileIDNormPCL = c_strID + "_02norm.pcl"
c_strFileIDPCL = c_strID + ".pcl"
c_strFileIDDAB = c_strID + ".dab"
c_strProgMergeTables = arepa.d(arepa.path_arepa(), arepa.c_strDirSrc,
                               "merge_tables.py")
c_strProgSamples2PCL = arepa.d(arepa.path_repo(), arepa.c_strDirSrc,
                               "samples2pcl.py")
c_afileInputSamples = Glob("../*sample_table*")

#===============================================================================
# Clip sample tables to first two columns
#===============================================================================
示例#6
0
if "testing" in locals():
    sys.exit()

pE = DefaultEnvironment()

c_strID = arepa.cwd()
c_fileInputC = sfle.d(pE, arepa.path_repo(), sfle.c_strDirTmp, "stringc")

c_fileIDPKL = sfle.d(pE, c_strID + ".pkl")
c_fileIDDAB = sfle.d(pE, c_strID + ".dab")
c_fileIDQUANT = sfle.d(c_strID + ".quant")
c_fileIDRawDAT = sfle.d(pE, c_strID + "_00raw.dat")
c_fileIDDAT = sfle.d(pE, c_strID + ".dat")

c_fileProgUnpickle = sfle.d(pE, arepa.path_arepa(), sfle.c_strDirSrc,
                            "unpickle.py")
c_fileProgC2Metadata = sfle.d(pE, arepa.path_repo(), sfle.c_strDirSrc,
                              "c2metadata.py")
c_fileProgC2DAT = sfle.d(pE, arepa.path_repo(), sfle.c_strDirSrc, "c2dat.py")
c_fileInputManCurTXT = sfle.d(pE, arepa.path_repo(), sfle.c_strDirEtc,
                              "manual_curation/", c_strID + sfle.c_strSufTXT)

c_fileInputSConscriptGM = sfle.d(pE, arepa.path_arepa(), sfle.c_strDirSrc,
                                 "SConscript_genemapping.py")
c_fileInputSConscriptDAB = sfle.d(pE, arepa.path_arepa(), sfle.c_strDirSrc,
                                  "SConscript_dat-dab.py")

c_fileStatus = sfle.d(pE, "status.txt")
c_strGeneFrom = None  # we do not know a priori what the gene identifiers are going to be