def load_previous(agp_file, nodes): ''' loads info from previous AGP''' # create node lookup. lookup = create_lookup(nodes) # load the agp array. agp_edges = load_agps(agp_file) # ensure sorted by scaffname and scafidx. agp_edges.sort(order=['scaf_name','scaf_idx']) # build list of component offsets. orien = dict() offsets = dict() for i in range(agp_edges.size): # skip non contigs. if agp_edges[i]['comp_type'] != "W": continue # save orientation. orien[lookup[agp_edges[i]['comp_name']]] = agp_edges[i]['comp_orien'] # record index. if agp_edges[i]['scaf_name'] not in offsets: offsets[agp_edges[i]['scaf_name']] = list() offsets[agp_edges[i]['scaf_name']].append(i) # add bundle info to this. gaps = dict() active = set() for key in offsets: # loop over edges. for i in range(len(offsets[key]) - 1): # get AGP edge. ea = agp_edges[offsets[key][i]] eb = agp_edges[offsets[key][i+1]] # get index. idxa = lookup[ea['comp_name']] idxb = lookup[eb['comp_name']] # get gap. gaps[(idxa,idxb)] = eb['scaf_start'] - ea['scaf_stop'] # note its active. active.add((idxa,idxb)) # return gaps, active set and node set. return gaps
def call_agp_gaps(agp_file, nodes): ''' calls agp gaps''' # create node lookup. lookup = create_lookup(nodes) # load the agp array. agp_edges = load_agps(agp_file) # ensure sorted by scaffname and scafidx. agp_edges.sort(order=['scaf_name','scaf_idx']) # build list of component offsets. offsets = dict() for i in range(agp_edges.size): # skip non contigs. if agp_edges[i]['comp_type'] != "W": continue # record index. if agp_edges[i]['scaf_name'] not in offsets: offsets[agp_edges[i]['scaf_name']] = list() offsets[agp_edges[i]['scaf_name']].append(i) # add bundle info to this. gaps = dict() for key in offsets: # loop over edges. for i in range(len(offsets[key]) - 1): # get AGP edge. ea = agp_edges[offsets[key][i]] eb = agp_edges[offsets[key][i+1]] # get index. idxa = lookup[ea['comp_name']] idxb = lookup[eb['comp_name']] # get gap. gaps[(idxa,idxb)] = eb['scaf_start'] - ea['scaf_stop'] return gaps
def make_key(a, b):
''' makes sorted key'''
if a < b:
return (a,b)
else:
return (b,a)
########### script ##################
# load hdf5 information.
logging.info("loading data arrays")
nodes = load_nodes(input_nodes_file)
edges = load_edges(input_edges_file)
agps = load_agps(input_agp_file)
nlookup = create_lookup(nodes)
# build bundle count.
logging.info("counting bundles")
blookup = dict()
for i in range(edges.size):
# get id
idxa = edges[i]['ctg_a_idx']
idxb = edges[i]['ctg_b_idx']
key = make_key(idxa, idxb)
# count it.
if key not in blookup:
blookup[key] = 0
blookup[key] += 1
