def main(args):
file_species_pairs = []
i = 1
while i < len(args)-2:
file_species_pairs.append([args[i], args[i+1]])
i += 2
cutoff = float(sys.argv[-2])
plot_dir = sys.argv[-1]
species2elms = {}
virus2elms = {}
# first grab virus ELMs
for file, species in file_species_pairs:
if file.find('flu') != -1:
virus2elms[species] = utils.get_seq2count_dict(file, cutoff)
else:
species2elms[species] = True
elms = {}
for species in virus2elms:
for elm in virus2elms[species]:
elms[elm] = True
for species in species2elms:
species2elms[species] = utils.get_seq2count_dict_for_seqs(file,
cutoff,
virus2elms)
for virus in virus2elms:
species2elms[virus] = virus2elms[virus]
for elm in elms:
if utils.check_ones(species2elms, elm):
if utils_distance.distance_elms(species2elms['Sus_scrofa'][elm],
species2elms['H_sapiens'][elm]) > float(-1) or utils_distance.distance_elms(species2elms['Sus_scrofa'][elm],
species2elms['Gallus_gallus'][elm]) > float(0):
utils_plot.elm_host_barplot(species2elms, elm,
os.path.join(plot_dir,
elm + '.virus_hosts.png'))
virus2dict['human'] = utils.get_seq2count_dict('results/flu_elmdict_human',
float(0.05))
genomes = ('H_sapiens', 'Gallus_gallus', 'Sus_scrofa')
species2dict = {}
for g in genomes:
species2dict[g] = utils.get_seq2count_dict('results/elmdict_'
+ g + '.txt',
float(0.01))
use_elms = {}
for elm in virus2dict['human']:
if elm in virus2dict['chicken'] and elm in virus2dict['swine']:
distance_is_0 = False
for v1, v2 in itertools.combinations(virus2dict.keys(), 2):
distance = utils_distance.distance_elms(virus2dict[v1][elm],
virus2dict[v2][elm])
if distance == float(0):
distance_is_0 = True
if not distance_is_0:
use_elms[elm] = True
host_elms = {}
for elm in species2dict['H_sapiens']:
if elm in species2dict['Gallus_gallus'] and elm in species2dict['Sus_scrofa']:
distance_is_0 = False
for h1, h2 in itertools.combinations(genomes, 2):
distance = utils_distance.distance_elms(species2dict[h1][elm],
species2dict[h2][elm])
if distance == float(0):
distance_is_0 = True
if not distance_is_0: