def starting_esrna(in_put, out_put, args_muscle, args_muscle_p, args_clustalw,
args_clustalw_p, args_gblocks, args_raxml, args_raxml_p,
args_fasttree, args_fasttree_p, args_thread,
args_extenddata):
'''reconstruct phylogenetic tree by ssu rna extend method'''
extend_check = checkFile(args_extenddata)
ssu_input = checkSilvaOrganism(in_put)
out_retrieve = retrieve16srna(ssu_input, out_put)
retrieve_srna_path = os.path.join(out_retrieve, 'rna_sequence.fasta')
fw = open(retrieve_srna_path, 'ab')
with open(extend_check) as read:
for line in read:
fw.write(line)
fw.close()
# set default aligned by muscle if not specify clustalw
if args_clustalw:
out_alg = doclustalw(out_retrieve, out_put, args_clustalw_p)
elif args_muscle:
out_alg = domuscle(out_retrieve, out_put, args_muscle_p)
# gblocks
if args_gblocks is gblockspara_pro:
args_gblocks = gblockspara_dna
out_gblock = dogblocks(out_alg, args_gblocks)
out_f2p = fasta2phy(out_gblock)
# reconstruct tree
if args_fasttree:
args_fasttree_p_add = "-nt " + args_fasttree_p.lstrip()
doFastTree(out_f2p, out_put, args_fasttree_p_add, args_thread)
elif args_raxml:
if args_raxml_p is raxmlpara_pro:
args_raxml_p = raxmlpara_dna
doraxml(out_f2p, out_put, args_raxml_p, args_thread)
def starting_esrna(in_put, out_put, args_muscle, args_muscle_p, args_clustalw,
args_clustalw_p, args_mafft, args_mafft_p, args_gblocks,
args_gblocks_p, args_trimal, args_trimal_p, args_raxml,
args_raxml_p, args_fasttree, args_fasttree_p, args_iqtree,
args_iqtree_p, args_thread, args_extenddata, args_db):
'''reconstruct phylogenetic tree by ssu rna extend method'''
extend_check = checkFile(args_extenddata)
ssu_input, recovery_dic = checkSilvaOrganism(in_put)
start = time.time()
out_retrieve = retrieve16srna(ssu_input, out_put, args_db)
end = time.time()
auto_build_log.info(
'Retrieving SSU rRNA sequences used time: {} Seconds'.format(end -
start))
if not recovery_dic == []:
recovery_silva(out_retrieve, recovery_dic, ssu_input)
retrieve_srna_path = os.path.join(out_retrieve, 'rna_sequence.fasta')
fw = open(retrieve_srna_path, 'a')
with open(extend_check) as read:
for line in read:
fw.write(line)
fw.close()
# set default aligned by muscle if not specify clustalw
start2 = time.time()
if args_clustalw:
out_alg = doclustalw(out_retrieve, out_put, args_clustalw_p)
elif args_mafft:
out_alg = domafft(out_retrieve, out_put, args_mafft_p)
elif args_muscle:
out_alg = domuscle(out_retrieve, out_put, args_muscle_p)
# set default trim by gblocks if not specify trimal
if args_trimal:
out_f2p = dotrimal(out_alg, args_trimal_p)
elif args_gblocks:
if args_gblocks_p is gblockspara_pro:
args_gblocks_p = gblockspara_dna
out_gblock = dogblocks(out_alg, args_gblocks_p)
out_f2p = fasta2phy(out_gblock)
# reconstruct tree
if args_fasttree:
args_fasttree_p_add = "-nt " + args_fasttree_p.lstrip()
doFastTree(out_f2p, out_put, args_fasttree_p_add, args_thread)
elif args_iqtree:
doiqtree(out_f2p, out_put, args_iqtree_p, args_thread)
elif args_raxml:
if args_raxml_p is raxmlpara_pro:
args_raxml_p = raxmlpara_dna
doraxml(out_f2p, out_put, args_raxml_p, args_thread)
end2 = time.time()
auto_build_log.info(
'Constructing species tree used time: {} Seconds'.format(end2 -
start2))
def starting_esrna(in_put, out_put,
args_muscle, args_muscle_p,
args_clustalw, args_clustalw_p,
args_mafft, args_mafft_p,
args_gblocks, args_gblocks_p,
args_trimal, args_trimal_p,
args_raxml, args_raxml_p,
args_fasttree, args_fasttree_p,
args_iqtree,args_iqtree_p,
args_thread, args_extenddata):
'''reconstruct phylogenetic tree by ssu rna extend method'''
extend_check = checkFile(args_extenddata)
ssu_input,recovery_dic = checkSilvaOrganism(in_put)
out_retrieve = retrieve16srna(ssu_input, out_put)
if not recovery_dic == {}:
recovery(out_retrieve,recovery_dic)
retrieve_srna_path = os.path.join(out_retrieve, 'rna_sequence.fasta')
fw = open(retrieve_srna_path, 'ab')
with open(extend_check) as read:
for line in read:
fw.write(line)
fw.close()
# set default aligned by muscle if not specify clustalw
if args_clustalw:
out_alg = doclustalw(out_retrieve, out_put, args_clustalw_p)
elif args_mafft:
out_alg = domafft(out_retrieve, out_put, args_mafft_p)
elif args_muscle:
out_alg = domuscle(out_retrieve, out_put, args_muscle_p)
# set default trim by gblocks if not specify trimal
if args_trimal:
out_f2p = dotrimal(out_alg, args_trimal_p)
elif args_gblocks:
if args_gblocks_p is gblockspara_pro:
args_gblocks_p = gblockspara_dna
out_gblock = dogblocks(out_alg, args_gblocks_p)
out_f2p = fasta2phy(out_gblock)
# reconstruct tree
if args_fasttree:
args_fasttree_p_add = "-nt " + args_fasttree_p.lstrip()
doFastTree(out_f2p, out_put, args_fasttree_p_add, args_thread)
elif args_iqtree:
doiqtree(out_f2p, out_put, args_iqtree_p, args_thread)
elif args_raxml:
if args_raxml_p is raxmlpara_pro:
args_raxml_p = raxmlpara_dna
doraxml(out_f2p, out_put, args_raxml_p, args_thread)
def starting_srna(in_put, out_put,
args_muscle, args_muscle_p,
args_clustalw, args_clustalw_p,
args_mafft, args_mafft_p,
args_gblocks, args_gblocks_p,
args_trimal, args_trimal_p,
args_raxml, args_raxml_p,
args_fasttree, args_fasttree_p,
args_iqtree,args_iqtree_p,
args_thread):
'''reconstruct phylogenetic tree by ssu rna method'''
ssu_input,recovery_dic = checkSilvaOrganism(in_put)
out_retrieve = retrieve16srna(ssu_input, out_put)
if not recovery_dic == {}:
recovery(out_retrieve,recovery_dic)
# set default aligned by muscle if not specify clustalw or mafft
if args_clustalw:
out_alg = doclustalw(out_retrieve, out_put, args_clustalw_p)
elif args_mafft:
out_alg = domafft(out_retrieve, out_put, args_mafft_p)
elif args_muscle:
out_alg = domuscle(out_retrieve, out_put, args_muscle_p)
# set default trim by gblocks if not specify trimal
if args_trimal:
out_f2p = dotrimal(out_alg, args_trimal_p)
elif args_gblocks:
if args_gblocks_p is gblockspara_pro:
args_gblocks_p = gblockspara_dna
out_gblock = dogblocks(out_alg, args_gblocks_p)
out_f2p = fasta2phy(out_gblock)
# reconstruct tree
if args_fasttree:
args_fasttree_p_add = "-nt " + args_fasttree_p.lstrip()
doFastTree(out_f2p, out_put, args_fasttree_p_add, args_thread)
elif args_iqtree:
doiqtree(out_f2p, out_put, args_iqtree_p, args_thread)
elif args_raxml:
if args_raxml_p is raxmlpara_pro:
args_raxml_p = raxmlpara_dna
doraxml(out_f2p, out_put, args_raxml_p, args_thread)
def build_srna(in_put, out_put, args_muscle, args_muscle_p, args_clustalw, args_clustalw_p,
args_gblocks, args_raxml, args_raxml_p, args_fasttree, args_fasttree_p, args_thread):
'''reconstruct phylogenetic tree by ssu rna method'''
out_retrieve = in_put
# set default aligned by muscle if not specify clustalw
if args_clustalw:
out_alg = doclustalw(out_retrieve, out_put, args_clustalw_p)
elif args_muscle:
out_alg = domuscle(out_retrieve, out_put, args_muscle_p)
if args_gblocks == gblockspara_pro:
args_gblocks = gblockspara_dna
out_gblock = dogblocks(out_alg, args_gblocks)
out_f2p = fasta2phy(out_gblock)
# reconstruct tree
if args_fasttree:
args_fasttree_p_add = "-nt " + args_fasttree_p.lstrip()
doFastTree(out_f2p, out_put, args_fasttree_p_add, args_thread)
elif args_raxml:
if args_raxml_p == raxmlpara_pro:
args_raxml_p = raxmlpara_dna
doraxml(out_f2p, out_put, args_raxml_p, args_thread)
def build_srna(in_put, out_put,
args_muscle, args_muscle_p,
args_clustalw, args_clustalw_p,
args_mafft, args_mafft_p,
args_gblocks, args_gblocks_p,
args_trimal, args_trimal_p,
args_raxml, args_raxml_p,
args_fasttree, args_fasttree_p,
args_iqtree,args_iqtree_p,
args_thread):
'''reconstruct phylogenetic tree by ssu rna method'''
out_retrieve = in_put
# set default aligned by muscle if not specify clustalw or mafft
if args_clustalw:
out_alg = doclustalw(out_retrieve, out_put, args_clustalw_p)
elif args_mafft:
out_alg = domafft(out_retrieve, out_put, args_mafft_p)
elif args_muscle:
out_alg = domuscle(out_retrieve, out_put, args_muscle_p)
# set default trim by gblocks if not specify trimal
if args_trimal:
out_f2p = dotrimal(out_alg, args_trimal_p)
elif args_gblocks:
if args_gblocks_p == gblockspara_pro:
args_gblocks_p = gblockspara_dna
out_gblock = dogblocks(out_alg, args_gblocks_p)
out_f2p = fasta2phy(out_gblock)
# reconstruct tree
if args_fasttree:
args_fasttree_p_add = "-nt " + args_fasttree_p.lstrip()
doFastTree(out_f2p, out_put, args_fasttree_p_add, args_thread)
elif args_iqtree:
doiqtree(out_f2p, out_put, args_iqtree_p, args_thread)
elif args_raxml:
if args_raxml_p == raxmlpara_pro:
args_raxml_p = raxmlpara_dna
doraxml(out_f2p, out_put, args_raxml_p, args_thread)
def build_srna(in_put, out_put, args_muscle, args_muscle_p, args_clustalw,
args_clustalw_p, args_mafft, args_mafft_p, args_gblocks,
args_gblocks_p, args_trimal, args_trimal_p, args_raxml,
args_raxml_p, args_fasttree, args_fasttree_p, args_iqtree,
args_iqtree_p, args_thread):
'''reconstruct phylogenetic tree by ssu rna method'''
start = time.time()
out_retrieve = in_put
# set default aligned by muscle if not specify clustalw or mafft
if args_clustalw:
out_alg = doclustalw(out_retrieve, out_put, args_clustalw_p)
elif args_mafft:
out_alg = domafft(out_retrieve, out_put, args_mafft_p)
elif args_muscle:
out_alg = domuscle(out_retrieve, out_put, args_muscle_p)
# set default trim by gblocks if not specify trimal
if args_trimal:
out_f2p = dotrimal(out_alg, args_trimal_p)
elif args_gblocks:
if args_gblocks_p == gblockspara_pro:
args_gblocks_p = gblockspara_dna
out_gblock = dogblocks(out_alg, args_gblocks_p)
out_f2p = fasta2phy(out_gblock)
# reconstruct tree
if args_fasttree:
args_fasttree_p_add = "-nt " + args_fasttree_p.lstrip()
doFastTree(out_f2p, out_put, args_fasttree_p_add, args_thread)
elif args_iqtree:
doiqtree(out_f2p, out_put, args_iqtree_p, args_thread)
elif args_raxml:
if args_raxml_p == raxmlpara_pro:
args_raxml_p = raxmlpara_dna
doraxml(out_f2p, out_put, args_raxml_p, args_thread)
end = time.time()
build_log.info(
'Constructing species tree used time: {} Seconds'.format(end - start))
def starting_srna(in_put, out_put, args_muscle, args_muscle_p, args_clustalw,
args_clustalw_p, args_gblocks, args_raxml, args_raxml_p,
args_fasttree, args_fasttree_p, args_thread):
'''reconstruct phylogenetic tree by ssu rna method'''
ssu_input = checkSilvaOrganism(in_put)
out_retrieve = retrieve16srna(ssu_input, out_put)
# set default aligned by muscle if not specify clustalw
if args_clustalw:
out_alg = doclustalw(out_retrieve, out_put, args_clustalw_p)
elif args_muscle:
out_alg = domuscle(out_retrieve, out_put, args_muscle_p)
# gblocks
if args_gblocks is gblockspara_pro:
args_gblocks = gblockspara_dna
out_gblock = dogblocks(out_alg, args_gblocks)
out_f2p = fasta2phy(out_gblock)
# reconstruct tree
if args_fasttree:
args_fasttree_p_add = "-nt " + args_fasttree_p.lstrip()
doFastTree(out_f2p, out_put, args_fasttree_p_add, args_thread)
elif args_raxml:
if args_raxml_p is raxmlpara_pro:
args_raxml_p = raxmlpara_dna
doraxml(out_f2p, out_put, args_raxml_p, args_thread)
from physpetool.phylotree.domuscle import domuscle, domuscle_file # test domuscle fuction # input = "/home/yangfang/physpetools/testdata/protein.fastq" # output = "/home/yangfang/physpetools/testdata/protein_alignment2.fasta" # domuscle(input, output) # test domuscle_file fuction # input = "/home/yangfang/physpetools/testdata/temp/conserved_protein" # output = "/home/yangfang/physpetools/testdata/phytree/test" # out_alg = domuscle_file(input, output) # print out_alg musclepara = '-maxiters 100' input = "/home/yangfang/physpetools/temp/16srnadata20160725152715" output = "/home/yangfang/physpetools/testdata/test16srandata" out_alg = domuscle(input, output, musclepara) print out_alg
from physpetool.phylotree.domuscle import domuscle, domuscle_file # test domuscle fuction # input = "/home/yangfang/physpetools/testdata/protein.fastq" # output = "/home/yangfang/physpetools/testdata/protein_alignment2.fasta" # domuscle(input, output) # test domuscle_file fuction # input = "/home/yangfang/physpetools/testdata/temp/conserved_protein" # output = "/home/yangfang/physpetools/testdata/phytree/test" # out_alg = domuscle_file(input, output) # print out_alg musclepara = '-maxiters 100' input = "/home/yangfang/physpetools/temp/16srnadata20160725152715" output = "/home/yangfang/physpetools/testdata/test16srandata" out_alg = domuscle(input, output,musclepara) print (out_alg)