def s04_novo_3_HTSeq_unknown(self):
sh_file = "%s/s04.novo_3.HTSeq_unknown.sh" % self.scripts
sh_work_file = "%s/s04.novo_3.HTSeq_unknown_work.sh" % self.scripts
l_sh_info = self.s_04_3_novo_HTSeq_unknown()
l_sh_work = []
dir_db = "%s/%s" % (self.Database, self.ref)
unknown_GTF = "%s/novo_lnc_raw_%s.combined.gtf" % (dir_db, self.s_idx)
for brief_name in self.samInfo_pd_RNA['brief_name']:
m_01.make_dir([self.dir_HTS_unknown, brief_name])
l_sh_work.append("sh %s %s %s" %
(sh_file, brief_name, unknown_GTF))
my_job = m_jobs.run_jobs(sh_file, sh_work_file, l_sh_info, l_sh_work)
my_job.running_multi(cpu=8, is_debug=self.is_debug)
# my_job.running_SGE(vf="400m", maxjob=100, is_debug = self.is_debug)
m_01.make_dir([self.dir_HTS_result])
Gtf_Info = m_gtf.GTFFeature(unknown_GTF)
Gtf_Info.gene_intergenic(self.intragenic_bed)
Cnt_Info = m_cnt.CountInfo(self.dir_HTS_unknown,
self.samInfo_pd_RNA['brief_name'],
"dexseq_NeoRaw", self.dir_HTS_result)
Cnt_Info.generate_mat()
Cnt_Info.load_mat()
Cnt_Info.cal_RPKM(Gtf_Info.gene, self.dir_tophat)
rpkm_file = "%s/merge.%s.RPKM.xls" % (self.dir_HTS_result,
"dexseq_NeoRaw")
Gtf_Info.load_gene_RPKM(rpkm_file)
Gtf_Info.output_GTF()
Gtf_Info.get_gene_info()
def __get_ref_index(self):
sh_file = "%s/db02.RefIndex.sh" % (self.scripts)
sh_work_file = "%s/db02.RefIndex_work.sh" % (self.scripts)
l_sh_info = self.db_02_BuildRefIndex()
l_sh_work = []
l_sh_work.append("sh %s %s" % (sh_file, self.ref))
my_job = m_jobs.run_jobs(sh_file, sh_work_file, l_sh_info, l_sh_work)
my_job.running_multi(cpu=8, is_debug=self.is_debug)
def __get_ref_fasta(self):
sh_file = "%s/db01.DownloadRef.sh" % (self.scripts)
sh_work_file = "%s/db01.DownloadRef_work.sh" % (self.scripts)
l_sh_info = self.db_01_DownloadRef()
l_sh_work = []
l_sh_work.append("sh %s %s" % (sh_file, self.ref))
my_job = m_jobs.run_jobs(sh_file, sh_work_file, l_sh_info, l_sh_work)
my_job.running_multi(cpu=8, is_debug=self.is_debug)
def __get_transcriptome(self):
sh_file = "%s/db05.transcriptome.sh" % (self.scripts)
sh_work_file = "%s/db05.transcriptome_work.sh" % (self.scripts)
l_sh_info = self.db_05_Transcriptome()
l_sh_work = []
l_sh_work.append("sh %s %s" % (sh_file, self.ref))
my_job = m_jobs.run_jobs(sh_file, sh_work_file, l_sh_info, l_sh_work)
my_job.running_multi(cpu=8, is_debug=self.is_debug)
def __get_rmsk(self):
sh_file = "%s/db04.rmsk.sh" % (self.scripts)
sh_work_file = "%s/db04.rmsk_work.sh" % (self.scripts)
l_sh_info = self.db_04_rmsk()
l_sh_work = []
l_sh_work.append("sh %s %s" % (sh_file, self.ref))
my_job = m_jobs.run_jobs(sh_file, sh_work_file, l_sh_info, l_sh_work)
my_job.running_multi(cpu=8, is_debug=self.is_debug)
def __get_ref_fasta(self):
sh_file = "%s/db01.DownloadRef.sh" % (self.scripts)
sh_work_file = "%s/db01.DownloadRef_work.sh" % (self.scripts)
l_sh_info = self.db_01_DownloadRef()
l_sh_work = []
l_sh_work.append("sh %s %s" % (sh_file,self.ref))
my_job = m_jobs.run_jobs(sh_file, sh_work_file, l_sh_info, l_sh_work)
my_job.running_multi(cpu=8, is_debug = self.is_debug)
def __get_transcriptome(self):
sh_file = "%s/db05.transcriptome.sh" % (self.scripts)
sh_work_file = "%s/db05.transcriptome_work.sh" % (self.scripts)
l_sh_info = self.db_05_Transcriptome()
l_sh_work = []
l_sh_work.append("sh %s %s" % (sh_file,self.ref))
my_job = m_jobs.run_jobs(sh_file, sh_work_file, l_sh_info, l_sh_work)
my_job.running_multi(cpu=8, is_debug = self.is_debug)
def __get_rmsk(self):
sh_file = "%s/db04.rmsk.sh" % (self.scripts)
sh_work_file = "%s/db04.rmsk_work.sh" % (self.scripts)
l_sh_info = self.db_04_rmsk()
l_sh_work = []
l_sh_work.append("sh %s %s" % (sh_file,self.ref))
my_job = m_jobs.run_jobs(sh_file, sh_work_file, l_sh_info, l_sh_work)
my_job.running_multi(cpu=8, is_debug = self.is_debug)
def __get_ref_index(self):
sh_file = "%s/db02.RefIndex.sh" % (self.scripts)
sh_work_file = "%s/db02.RefIndex_work.sh" % (self.scripts)
l_sh_info = self.db_02_BuildRefIndex()
l_sh_work = []
l_sh_work.append("sh %s %s" % (sh_file,self.ref))
my_job = m_jobs.run_jobs(sh_file, sh_work_file, l_sh_info, l_sh_work)
my_job.running_multi(cpu=8, is_debug = self.is_debug)
def __get_HTS_clean_split(self):
sh_file = "%s/stat02.SplitHTS.sh" % (self.scripts)
sh_work_file = "%s/stat02.SplitHTS_work.sh" % (self.scripts)
l_sh_info = self.stat_02_SplitHTS()
l_sh_work = []
l_sh_work.append("sh %s %s" % (sh_file, self.use_gtf))
my_job = m_jobs.run_jobs(sh_file, sh_work_file, l_sh_info, l_sh_work)
my_job.running_multi(cpu=8, is_debug=self.is_debug)
def s05_2_extra_makeGTF(self, given_GTF, given_GTF_ERCC):
sh_file = "%s/s05.2.GTF_ercc.sh" % (self.scripts)
sh_work_file = "%s/s05.2.GTF_ercc_work.sh" % (self.scripts)
l_sh_info = self.s_05_2_erccGTF(given_GTF, given_GTF_ERCC)
l_sh_work = []
l_sh_work.append("sh %s %s" % (sh_file, self.ref))
my_job = m_jobs.run_jobs(sh_file, sh_work_file, l_sh_info, l_sh_work)
my_job.running_multi(cpu=8, is_debug = self.is_debug)
def __get_HTS_clean_split(self):
sh_file = "%s/stat02.SplitHTS.sh" % (self.scripts)
sh_work_file = "%s/stat02.SplitHTS_work.sh" % (self.scripts)
l_sh_info = self.stat_02_SplitHTS()
l_sh_work = []
l_sh_work.append("sh %s %s" % (sh_file, self.use_gtf))
my_job = m_jobs.run_jobs(sh_file, sh_work_file, l_sh_info, l_sh_work)
my_job.running_multi(cpu=8, is_debug = self.is_debug)
def s05_2_extra_makeGTF(self, given_GTF, given_GTF_ERCC):
sh_file = "%s/s05.2.GTF_ercc.sh" % (self.scripts)
sh_work_file = "%s/s05.2.GTF_ercc_work.sh" % (self.scripts)
l_sh_info = self.s_05_2_erccGTF(given_GTF, given_GTF_ERCC)
l_sh_work = []
l_sh_work.append("sh %s %s" % (sh_file, self.ref))
my_job = m_jobs.run_jobs(sh_file, sh_work_file, l_sh_info, l_sh_work)
my_job.running_multi(cpu=8, is_debug=self.is_debug)
def s03_HTSeq_known(self):
sh_file = "%s/s03.HTSeq.sh" % (self.scripts)
sh_work_file = "%s/s03.HTSeq_work.sh" % (self.scripts)
l_sh_info = self.s_03_HTSeq_known()
l_sh_work = []
for brief_name in self.samInfo_pd_RNA['brief_name']:
m_01.make_dir([ self.dir_HTS_known, brief_name ])
l_sh_work.append("sh %s %s" % (sh_file, brief_name))
my_job = m_jobs.run_jobs(sh_file, sh_work_file, l_sh_info, l_sh_work)
my_job.running_multi(cpu=8, is_debug = self.is_debug)
def s03_HTSeq_known(self):
sh_file = "%s/s03.HTSeq.sh" % (self.scripts)
sh_work_file = "%s/s03.HTSeq_work.sh" % (self.scripts)
l_sh_info = self.s_03_HTSeq_known()
l_sh_work = []
for brief_name in self.samInfo_pd_RNA['brief_name']:
m_01.make_dir([self.dir_HTS_known, brief_name])
l_sh_work.append("sh %s %s" % (sh_file, brief_name))
my_job = m_jobs.run_jobs(sh_file, sh_work_file, l_sh_info, l_sh_work)
my_job.running_multi(cpu=8, is_debug=self.is_debug)
def s08_repeatCount(self):
sh_file = "%s/s08.RepCnt.sh" % (self.scripts)
sh_work_file = "%s/s08.RepCnt_work.sh" % (self.scripts)
l_sh_info = self.s_08_repeatCount()
l_sh_work = []
for brief_name in self.samInfo_pd_RNA['brief_name']:
m_01.make_dir([ self.dir_repeat_counts, brief_name ])
l_sh_work.append("sh %s %s" % (sh_file, brief_name))
my_job = m_jobs.run_jobs(sh_file, sh_work_file, l_sh_info, l_sh_work)
my_job.running_multi(cpu=8, is_debug = self.is_debug)
def s08_repeatCount(self):
sh_file = "%s/s08.RepCnt.sh" % (self.scripts)
sh_work_file = "%s/s08.RepCnt_work.sh" % (self.scripts)
l_sh_info = self.s_08_repeatCount()
l_sh_work = []
for brief_name in self.samInfo_pd_RNA['brief_name']:
m_01.make_dir([self.dir_repeat_counts, brief_name])
l_sh_work.append("sh %s %s" % (sh_file, brief_name))
my_job = m_jobs.run_jobs(sh_file, sh_work_file, l_sh_info, l_sh_work)
my_job.running_multi(cpu=8, is_debug=self.is_debug)
def s06_1_cufflinks(self, cflk_dir, use_gtf):
sh_file = "%s/s06.1.cflk.sh" % (self.scripts)
sh_work_file = "%s/s06.1.cflk_work.sh" % (self.scripts)
l_sh_info = self.s_06_1_cflk(cflk_dir)
l_sh_work = []
for brief_name in self.samInfo_pd_RNA['brief_name']:
m_01.make_dir([ cflk_dir, brief_name ])
l_sh_work.append("sh %s %s %s" % (sh_file, brief_name, use_gtf))
my_job = m_jobs.run_jobs(sh_file, sh_work_file, l_sh_info, l_sh_work)
my_job.running_multi(cpu=8, is_debug = self.is_debug)
def s06_1_cufflinks(self, cflk_dir, use_gtf):
sh_file = "%s/s06.1.cflk.sh" % (self.scripts)
sh_work_file = "%s/s06.1.cflk_work.sh" % (self.scripts)
l_sh_info = self.s_06_1_cflk(cflk_dir)
l_sh_work = []
for brief_name in self.samInfo_pd_RNA['brief_name']:
m_01.make_dir([cflk_dir, brief_name])
l_sh_work.append("sh %s %s %s" % (sh_file, brief_name, use_gtf))
my_job = m_jobs.run_jobs(sh_file, sh_work_file, l_sh_info, l_sh_work)
my_job.running_multi(cpu=8, is_debug=self.is_debug)
def s04_novo_1_cufflinks_u(self):
sh_file = "%s/s04.novo_1.cufflinks_unknown.sh" %\
(self.scripts)
sh_work_file = "%s/s04.novo_1.cufflinks_unknown_work.sh" %\
(self.scripts)
l_sh_info = self.s_04_1_novo_HTSeq_unknown()
l_sh_work = []
for brief_name in self.samInfo_pd_RNA['brief_name']:
m_01.make_dir([self.dir_cufflinks_unknown, brief_name])
l_sh_work.append("sh %s %s" % (sh_file, brief_name))
my_job = m_jobs.run_jobs(sh_file, sh_work_file, l_sh_info, l_sh_work)
my_job.running_multi(cpu=8, is_debug=self.is_debug)
def s04_novo_1_cufflinks_u(self):
sh_file = "%s/s04.novo_1.cufflinks_unknown.sh" %\
(self.scripts)
sh_work_file = "%s/s04.novo_1.cufflinks_unknown_work.sh" %\
(self.scripts)
l_sh_info = self.s_04_1_novo_HTSeq_unknown()
l_sh_work = []
for brief_name in self.samInfo_pd_RNA['brief_name']:
m_01.make_dir([ self.dir_cufflinks_unknown, brief_name ])
l_sh_work.append("sh %s %s" % (sh_file, brief_name))
my_job = m_jobs.run_jobs(sh_file, sh_work_file, l_sh_info, l_sh_work)
my_job.running_multi(cpu=8, is_debug = self.is_debug)
def s01_QC(self, core_num=4):
sh_file = "%s/s01.QC.sh" % (self.scripts)
sh_work_file = "%s/s01.QC_work.sh" % (self.scripts)
l_sh_info = self.s_01_QC()
l_sh_work = []
for samp in self.samInfo_pd_RNA['sample']:
make_dir([self.dir_clean_data, samp])
idx = (self.samInfo_pd_RNA['sample'] == samp)
end = self.samInfo_pd_RNA[idx]['end_type'].values[0]
data_dype = self.M_CvtEnd[end]
l_sh_work.append("sh %s %s %d" % (sh_file, samp, data_dype))
my_job = m_jobs.run_jobs(sh_file, sh_work_file, l_sh_info, l_sh_work)
my_job.running_multi(cpu=core_num, is_debug=self.is_debug)
def s01_QC(self):
sh_file = "%s/s01.QC.sh" % (self.scripts)
sh_work_file = "%s/s01.QC_work.sh" % (self.scripts)
l_sh_info = self.s_01_QC()
l_sh_work = []
for samp in self.samInfo_pd_RNA['sample']:
make_dir( [ self.dir_clean_data, samp ] )
idx = (self.samInfo_pd_RNA['sample'] == samp)
end = self.samInfo_pd_RNA[ idx ]['end_type'].values[0]
data_dype = self.M_CvtEnd[ end ]
l_sh_work.append("sh %s %s %d" % (sh_file, samp, data_dype) )
my_job = m_jobs.run_jobs(sh_file, sh_work_file, l_sh_info, l_sh_work)
my_job.running_multi(cpu=8, is_debug = self.is_debug)
def s05_1_extra_makeGTF(self, extra_GTF=None):
sh_file = "%s/s05.1.mergeGTF.sh" % (self.scripts)
sh_work_file = "%s/s05.1.mergeGTF_work.sh" % (self.scripts)
l_sh_info = self.s_05_1_mergeGTF()
l_sh_work = []
dir_db = "%s/%s" % (self.Database, self.ref)
tail_str = "combined.FPKM0.5_rep0.25.multiExon.gtf"
unknown_GTF = "%s/novo_lnc_raw_%s.%s" % (dir_db, self.s_idx, tail_str)
l_sh_work.append("sh %s %s %s" % (sh_file, unknown_GTF, extra_GTF))
my_job = m_jobs.run_jobs(sh_file, sh_work_file, l_sh_info, l_sh_work)
my_job.running_multi(cpu=8, is_debug = self.is_debug)
def s05_1_extra_makeGTF(self, extra_GTF=None):
sh_file = "%s/s05.1.mergeGTF.sh" % (self.scripts)
sh_work_file = "%s/s05.1.mergeGTF_work.sh" % (self.scripts)
l_sh_info = self.s_05_1_mergeGTF()
l_sh_work = []
dir_db = "%s/%s" % (self.Database, self.ref)
tail_str = "combined.FPKM0.5_rep0.25.multiExon.gtf"
unknown_GTF = "%s/novo_lnc_raw_%s.%s" % (dir_db, self.s_idx, tail_str)
l_sh_work.append("sh %s %s %s" % (sh_file, unknown_GTF, extra_GTF))
my_job = m_jobs.run_jobs(sh_file, sh_work_file, l_sh_info, l_sh_work)
my_job.running_multi(cpu=8, is_debug=self.is_debug)
def s02_Tophat(self, core_num=1):
sh_file = "%s/s02.Tophat.sh" % (self.scripts)
sh_work_file = "%s/s02.Tophat_work.sh" % (self.scripts)
l_sh_info = self.s_02_Tophat()
l_sh_work = []
for samp in self.samInfo_pd_RNA['sample']:
idx = (self.samInfo_pd_RNA['sample'] == samp)
brief_name = self.samInfo_pd_RNA[idx]['brief_name'].values[0]
make_dir([self.dir_tophat, brief_name])
end = self.samInfo_pd_RNA[idx]['end_type'].values[0]
data_dype = self.M_CvtEnd[end]
l_sh_work.append("sh %s %s %s %d" %\
(sh_file, samp, brief_name, data_dype))
my_job = m_jobs.run_jobs(sh_file, sh_work_file, l_sh_info, l_sh_work)
my_job.running_multi(cpu=core_num, is_debug=self.is_debug)
def s02_Tophat(self, core_num=1):
sh_file = "%s/s02.Tophat.sh" % (self.scripts)
sh_work_file = "%s/s02.Tophat_work.sh" % (self.scripts)
l_sh_info = self.s_02_Tophat()
l_sh_work = []
for samp in self.samInfo_pd_RNA['sample']:
idx =(self.samInfo_pd_RNA['sample'] == samp)
brief_name= self.samInfo_pd_RNA[ idx ]['brief_name'].values[0]
make_dir( [ self.dir_tophat, brief_name ] )
end = self.samInfo_pd_RNA[ idx ]['end_type'].values[0]
data_dype = self.M_CvtEnd[ end ]
l_sh_work.append("sh %s %s %s %d" %\
(sh_file, samp, brief_name, data_dype))
my_job = m_jobs.run_jobs(sh_file, sh_work_file, l_sh_info, l_sh_work)
my_job.running_multi(cpu=core_num, is_debug = self.is_debug)
def s07_cuffnorm(self, cflk_dir, cfnm_dir, use_gtf):
sh_file = "%s/s07.cfnm.sh" % (self.scripts)
sh_work_file = "%s/s07.cfnm_work.sh" % (self.scripts)
l_sh_info = self.s_07_cfnm(cfnm_dir)
l_sh_work = []
m_01.make_dir([cfnm_dir])
l_brief = self.samInfo_pd_RNA['brief_name']
l_cxb = ["%s/%s/abundances.cxb" % (cflk_dir, sam) for sam in l_brief]
list_sam = ",".join(l_brief)
list_cxb = " ".join(l_cxb)
l_sh_work.append("sh %s %s %s %s" %
(sh_file, list_sam, use_gtf, list_cxb))
my_job = m_jobs.run_jobs(sh_file, sh_work_file, l_sh_info, l_sh_work)
my_job.running_multi(cpu=8, is_debug=self.is_debug)
def s04_novo_2_cuffcomp_trans(self):
sh_file = "%s/s04.novo_2.cuffcomp_unknown.sh" % self.scripts
sh_work_file = "%s/s04.novo_2.cuffcomp_unknown_work.sh" % self.scripts
l_sh_info = self.s_04_2_novo_cuffcomp()
l_sh_work = []
dir_db = "%s/%s" % (self.Database, self.ref)
out_prefix = "%s/novo_lnc_raw_%s" % (dir_db, self.s_idx)
l_input = []
for brief_name in self.samInfo_pd_RNA['brief_name']:
infile = "%s/%s/transcripts.gtf" %\
(self.dir_cufflinks_unknown, brief_name)
l_input.append(infile)
insam = " ".join(l_input)
l_sh_work.append("sh %s %s %s" % (sh_file, out_prefix, insam))
my_job = m_jobs.run_jobs(sh_file, sh_work_file, l_sh_info, l_sh_work)
my_job.running_multi(cpu=8, is_debug = self.is_debug)
def s07_cuffnorm(self, cflk_dir, cfnm_dir, use_gtf):
sh_file = "%s/s07.cfnm.sh" % (self.scripts)
sh_work_file = "%s/s07.cfnm_work.sh" % (self.scripts)
l_sh_info = self.s_07_cfnm(cfnm_dir)
l_sh_work = []
m_01.make_dir([ cfnm_dir ])
l_brief = self.samInfo_pd_RNA['brief_name']
l_cxb = [ "%s/%s/abundances.cxb" % (cflk_dir, sam) for sam in l_brief]
list_sam = ",".join(l_brief)
list_cxb = " ".join(l_cxb)
l_sh_work.append(
"sh %s %s %s %s" % (sh_file, list_sam, use_gtf, list_cxb)
)
my_job = m_jobs.run_jobs(sh_file, sh_work_file, l_sh_info, l_sh_work)
my_job.running_multi(cpu=8, is_debug = self.is_debug)
def s04_novo_2_cuffcomp_trans(self):
sh_file = "%s/s04.novo_2.cuffcomp_unknown.sh" % self.scripts
sh_work_file = "%s/s04.novo_2.cuffcomp_unknown_work.sh" % self.scripts
l_sh_info = self.s_04_2_novo_cuffcomp()
l_sh_work = []
dir_db = "%s/%s" % (self.Database, self.ref)
out_prefix = "%s/novo_lnc_raw_%s" % (dir_db, self.s_idx)
l_input = []
for brief_name in self.samInfo_pd_RNA['brief_name']:
infile = "%s/%s/transcripts.gtf" %\
(self.dir_cufflinks_unknown, brief_name)
l_input.append(infile)
insam = " ".join(l_input)
l_sh_work.append("sh %s %s %s" % (sh_file, out_prefix, insam))
my_job = m_jobs.run_jobs(sh_file, sh_work_file, l_sh_info, l_sh_work)
my_job.running_multi(cpu=8, is_debug=self.is_debug)
def s04_novo_3_HTSeq_unknown(self):
sh_file = "%s/s04.novo_3.HTSeq_unknown.sh" % self.scripts
sh_work_file = "%s/s04.novo_3.HTSeq_unknown_work.sh" % self.scripts
l_sh_info = self.s_04_3_novo_HTSeq_unknown()
l_sh_work = []
dir_db = "%s/%s" % (self.Database, self.ref)
unknown_GTF = "%s/novo_lnc_raw_%s.combined.gtf" % (dir_db, self.s_idx)
for brief_name in self.samInfo_pd_RNA['brief_name']:
m_01.make_dir([ self.dir_HTS_unknown, brief_name ])
l_sh_work.append("sh %s %s %s" %(sh_file, brief_name,unknown_GTF))
my_job = m_jobs.run_jobs(sh_file, sh_work_file, l_sh_info, l_sh_work)
my_job.running_multi(cpu=8, is_debug = self.is_debug)
# my_job.running_SGE(vf="400m", maxjob=100, is_debug = self.is_debug)
m_01.make_dir([ self.dir_HTS_result ])
Gtf_Info = m_gtf.GTFFeature( unknown_GTF )
Gtf_Info.gene_intergenic( self.intragenic_bed )
Cnt_Info = m_cnt.CountInfo(
self.dir_HTS_unknown,
self.samInfo_pd_RNA['brief_name'],
"dexseq_NeoRaw",
self.dir_HTS_result
)
Cnt_Info.generate_mat()
Cnt_Info.load_mat()
Cnt_Info.cal_RPKM(Gtf_Info.gene, self.dir_tophat)
rpkm_file = "%s/merge.%s.RPKM.xls" % ( self.dir_HTS_result,
"dexseq_NeoRaw"
)
Gtf_Info.load_gene_RPKM( rpkm_file )
Gtf_Info.output_GTF()
Gtf_Info.get_gene_info()
