def concat(genelistdict,
workdir_comb,
email,
num_threads=None,
percentage=0.37,
user_concat_fn=None,
backbone=False):
"""This is to concatenate different physcraper runs into a single alignment and tree.
genelistdict is a dict with gene names as key and the corresponding workdir
"""
license_print()
if not os.path.exists(path="{}/concat_checkpoint.p".format(workdir_comb)):
if not os.path.exists(
path="{}/load_single_data.p".format(workdir_comb)):
# save_copy_code(workdir_comb)
conc = Concat(workdir_comb, email)
conc.concatfile = user_concat_fn
for item in genelistdict.keys():
conc.load_single_genes(genelistdict[item]["workdir"],
genelistdict[item]["pickle"], item)
conc.combine()
else:
sys.stdout.write("load single data dump file\n")
conc = pickle.load(
open("{}/load_single_data.p".format(workdir_comb), "rb"))
# conc.dump()
conc.sp_seq_counter()
conc.get_largest_tre()
conc.make_sp_gene_dict()
conc.make_alns_dict()
conc.concatenate_alns()
conc.get_short_seq_from_concat(percentage)
conc.remove_short_seq()
conc.dump()
else:
sys.stdout.write("load concat_checkpoint dump file\n")
conc = pickle.load(
open("{}/concat_checkpoint.p".format(workdir_comb), "rb"))
conc.backbone = backbone
conc.make_concat_table()
conc.write_partition()
conc.write_otu_info()
conc.place_new_seqs(num_threads)
if backbone is False:
conc.calculate_bootstrap(num_threads)
conc.write_labelled('RAxML_bestTree.autoMRE_fa')
else:
conc.est_full_tree(num_threads)
conc.write_labelled('RAxML_bestTree.backbone_concat')
return conc
def test_run_raxml_concat():
time.sleep(10) # needs time so that file of order1 is written before
workdir_its = "tests/data/precooked/concat_pre"
workdir_ets = "tests/data/precooked/concat_pre"
email = "*****@*****.**"
pickle_fn = "final_ATT_checkpoint.p"
email = "*****@*****.**"
percentage = 0.4
num_threads = 2
workdir_comb = "./tests/output/concat_test"
genelistdict = {"its": {"workdir": workdir_its, "pickle": "its_{}".format(pickle_fn)},
"ets": {"workdir": workdir_ets, "pickle": "ets_{}".format(pickle_fn)}}
conc = Concat(workdir_comb, email)
for item in genelistdict.keys():
conc.load_single_genes(genelistdict[item]["workdir"], genelistdict[item]["pickle"], item)
conc.combine()
assert os.path.exists("{}/load_single_data.p".format(workdir_comb))
conc.sp_seq_counter()
conc.get_largest_tre()
assert conc.tre_start_gene == "ets"
conc.make_sp_gene_dict()
conc.make_alns_dict()
conc.concatenate_alns()
conc.get_short_seq_from_concat(percentage)
conc.remove_short_seq()
conc.backbone = True
conc.make_concat_table()
conc.write_partition()
conc.write_otu_info()
if conc.backbone is False:
conc.calculate_bootstrap(num_threads)
conc.write_labelled('RAxML_bestTree.autoMRE_fa')
else:
conc.est_full_tree(num_threads)
conc.write_labelled('RAxML_bestTree.backbone_concat')
conc.backbone = False
conc.place_new_seqs(num_threads)