def compute_overlaps(x,y): T1 = node.tree(y) T2 = node.tree(x) ct1 = 0 ct2 = 0 for start, stop in x: if T1.searchInterval((start, stop)): ct1+=1 for start, stop in y: if T2.searchInterval((start, stop)): ct2+=1 return ct1, ct2
def compute_overlaps(x,y,OUT=""): y.sort() T = node.tree(y) j,N,O = 0,len(y),list() #FHW.open(OUT,"w") for start, stop in x: FINDS = T.searchInterval((start, stop)) cx = (stop+start) / 2. for st,sp in FINDS: cy = (sp+st) / 2. O.append((cx-cy)) return O
def get_TSS_tree(FILE):
G = {}
with open(FILE) as FH:
for line in FH:
chrom, start, stop = line.strip("\n").split("\t")[:3]
start, stop = int(start), int(stop)
if chrom not in G:
G[chrom] = list()
G[chrom].append((start, stop))
for chrom in G:
G[chrom].sort()
G[chrom]=node.tree(G[chrom])
return G
def load_RNA_seq(FILE):
G = {}
R = {}
with open(FILE) as FH:
for line in FH:
gene, chrom, start, stop, cov = re.split("\s+", line.strip("\n"))[:5]
G[gene] = (chrom, start, stop, float(cov))
if chrom not in R:
R[chrom] = list()
R[chrom].append((int(start), int(stop), gene))
for chrom in R:
R[chrom] = node.tree(R[chrom])
return G, R
def run(N1,N2,T=100,l=100,alpha=1,beta=1):
A_stats = {}
B_stats = {}
AOs, BOs = list(),list()
for t in range(T):
A = [(x,x+alpha) for x in np.random.uniform(0, l-alpha, N1)]
B = [(x,x+beta) for x in np.random.uniform(0, l-beta, N2)]
A.sort()
B.sort()
TA = node.tree(A)
TB = node.tree(B)
AO,BO = 0,0
for a_st, a_sp in A:
FINDS = TB.searchInterval((a_st, a_sp))
if len(FINDS) not in A_stats:
A_stats[len(FINDS)] = 0
A_stats[len(FINDS)]+=1
AO+=len(FINDS)
for a_st, a_sp in B:
FINDS = TA.searchInterval((a_st, a_sp))
if len(FINDS) not in B_stats:
B_stats[len(FINDS)] = 0
B_stats[len(FINDS)]+=1
BO+=len(FINDS)
AOs.append(AO)
BOs.append(BO)
F = plt.figure(figsize=(15,10))
ax1 = F.add_subplot(2,2,1)
ax1.set_title("List A; N: " + str(N1) )
ax1.hist([b for b in A_stats],weights=np.array([A_stats[b] for b in A_stats]) / float(sum([A_stats[b] for b in A_stats])), alpha=0.3)
ax1.scatter([b for b in A_stats],[prob_single(b,alpha+beta, l,N2) for b in A_stats] )
ax1.set_xticks(range(0, max(A_stats.keys()) +1) )
ax2 = F.add_subplot(2,2,2)
ax2.set_title("List B; N: " + str(N2) )
ax2.hist([b for b in B_stats],weights=np.array([B_stats[b] for b in B_stats]) / float(sum([B_stats[b] for b in B_stats])), alpha=0.3)
ax2.set_xticks(range(0, max(B_stats.keys())+1))
ax2.set_xticklabels([str(i) for i in range(0,max(B_stats.keys())+1)])
ax2.scatter([b for b in B_stats],[prob_single(b,alpha+beta, l,N1) for b in B_stats] )
ax3 = F.add_subplot(2,2,3)
ax3.set_title("Total Number of Overlapping Events on A")
counts,edges = np.histogram(AOs,bins=max(AOs)-min(AOs))
edges = edges[:-1]
counts =[float(ct)/float(sum(counts)) for ct in counts]
ax3.bar(edges,counts , alpha=0.3)
#ax3.set_xticks(range(0, max(AOs) ) )
xs = np.linspace(min(AOs), max(AOs))
mu = np.mean(AOs)
std = np.std(AOs)
ax3.scatter(AOs, [ prob_single(a,alpha+beta, l,N2*N1) for a in AOs])
ax3.plot(xs,[ normal(x, mu, std) for x in xs])
ax4 = F.add_subplot(2,2,4)
ax4.set_title("Total Number of Overlapping Events on B")
counts,edges = np.histogram(BOs,bins=max(BOs)-min(BOs))
edges = edges[:-1]
counts =[float(ct)/float(sum(counts)) for ct in counts]
ax4.bar(edges,counts , alpha=0.3)
#ax3.set_xticks(range(0, max(AOs) ) )
xs = np.linspace(min(BOs), max(BOs))
mu = np.mean(BOs)
std = np.std(BOs)
ax4.scatter(BOs, [ prob_single(a,alpha+beta, l,N2*N1) for a in BOs])
ax4.plot(xs,[ normal(x, mu, std) for x in xs])
plt.show()