def make_ramp(strand, sign=1):
transfrags = []
chrom = 'chr1'
start = 1000
end = 1220
change_expr = 0.0
base_expr = 0.0
# "flat" part of expression landscape
expr = 1.0
for i in xrange(0, 50):
t = Transfrag(chrom=chrom, strand=strand,
_id='T1.%d' % i, sample_id='S%d' % i,
expr=expr, is_ref=False,
exons=[Exon(start, end)])
transfrags.append(t)
change_expr += expr
base_expr += expr
# "changing" area
i = 0
expr = 10.0
for pos in range(1100, 1120):
left, right = (start, pos) if sign < 0 else (pos, end)
t = Transfrag(chrom=chrom, strand=strand,
_id='T2.%d' % i, sample_id='S%d' % i,
expr=expr, is_ref=False,
exons=[Exon(left, right)])
transfrags.append(t)
change_expr += expr
i += 1
return chrom, start, end, strand, change_expr, base_expr, transfrags
def test_splices():
t = Transfrag(chrom='chrTest',
strand=Strand.POS,
exons=[Exon(0, 10), Exon(20, 30),
Exon(40, 50)])
splice_sites = frozenset(t.itersplices())
assert len(splice_sites) == 4
assert splice_sites == frozenset((10, 20, 30, 40))
def test_introns():
t = Transfrag(chrom='chrTest',
strand=Strand.POS,
exons=[Exon(0, 10), Exon(20, 30),
Exon(40, 50)])
introns = list(t.iterintrons())
assert len(introns) == 2
assert introns[0] == (10, 20)
assert introns[1] == (30, 40)
def test_trimming_to_zero_bug():
t_dict, locus = read_single_locus('change_point_bug.gtf')
transfrags_un = locus.get_transfrags(Strand.NA)
sgraph = SpliceGraph.create(transfrags_un)
cps = sgraph.detect_change_points()
for cp in cps:
sgraph.apply_change_point(cp)
sgraph.recreate()
# get start/stop nodes
start_nodes, stop_nodes = sgraph.get_start_stop_nodes()
# convert to node intervals
start_nodes = set(sgraph.get_node_interval(n_id) for n_id in start_nodes)
stop_nodes = set(sgraph.get_node_interval(n_id) for n_id in stop_nodes)
assert Exon(173433532, 173435169) in stop_nodes
assert Exon(173433532, 173435169) in start_nodes
assert Exon(173433532, 173435169) in start_nodes
def test_path_graph2():
return
t_dict, locus = read_single_locus('change_point2.gtf')
sgraph = SpliceGraph.create(t_dict.values())
# trivial case without additional stops or starts
k = 1
K = create_path_graph(sgraph, k)
kmer_id_map = K.graph['kmer_id_map']
n_id = sgraph.get_node_id(Exon(0, 100))
kmer_id = kmer_id_map[(n_id, )]
assert K.node[kmer_id]['expr'] == 12.0
assert K.node[SOURCE]['expr'] == 12.0
assert K.node[SINK]['expr'] == 12.0
# add a stop site
sgraph.stop_sites.add(50)
sgraph.recreate()
K = create_path_graph(sgraph, k=2)
kmer_id_map = K.graph['kmer_id_map']
n1 = (sgraph.get_node_id((0, 50)), sgraph.get_node_id((50, 100)))
kmer1 = kmer_id_map[n1]
n2 = (sgraph.get_node_id((0, 50)), )
kmer2 = kmer_id_map[n2]
assert K.node[kmer1]['expr'] == 1.0
assert K.node[kmer2]['expr'] == 10.0
assert K.node[SOURCE]['expr'] == 11.0
assert K.node[SINK]['expr'] == 11.0
# smooth kmer graph
smooth_graph(K)
assert K.node[kmer1]['expr'] == 1.0
assert K.node[kmer2]['expr'] == 10.0
assert K.node[SOURCE]['expr'] == 11.0
assert K.node[SINK]['expr'] == 11.0
# TODO: test after rescuing short transfrags
# add both a start and a stop site
sgraph.start_sites.add(50)
sgraph.stop_sites.add(50)
sgraph.recreate()
K = create_path_graph(sgraph, k=2)
smooth_graph(K)
kmer_id_map = K.graph['kmer_id_map']
n1 = (sgraph.get_node_id((0, 50)), sgraph.get_node_id((50, 100)))
n2 = (sgraph.get_node_id((0, 50)), )
n3 = (sgraph.get_node_id((50, 100)), )
kmer1 = kmer_id_map[n1]
kmer2 = kmer_id_map[n2]
kmer3 = kmer_id_map[n3]
assert K.node[kmer1]['expr'] == 1.0
assert K.node[kmer2]['expr'] == 10.0
assert K.node[kmer3]['expr'] == 1.0
assert K.node[SOURCE]['expr'] == 12.0
assert K.node[SINK]['expr'] == 12.0
def test_ccle55_cuff_noc2l():
'''Locus containing from 55 CCLE samples assembled with Cufflinks'''
# pull SpliceGraph out of GTF
t_dict, locus = read_single_locus('noc2l_locus.gtf')
found_sgraph = False
for sgraph in locus.create_splice_graphs():
if (sgraph.chrom == 'chr1' and sgraph.start == 934942 and
sgraph.end == 976702 and sgraph.strand == Strand.NEG):
found_sgraph = True
break
assert found_sgraph
# examine specific change points
trim = False
pval = 0.05
fc_cutoff = 0.8
n1 = Exon(934942, 944589)
n1_id = sgraph.get_node_id(n1)
assert sgraph.G.node[n1_id][SGNode.IS_STOP]
cps = sgraph.detect_change_points(pval=pval, fc_cutoff=fc_cutoff)
for cp in cps:
sgraph.apply_change_point(cp, trim=trim)
true_starts = set([964528, 957434, 959316])
true_stops = set([944278])
assert true_starts.issubset(sgraph.start_sites)
assert true_stops.issubset(sgraph.stop_sites)
# rebuild graph and examine start/stop nodes
sgraph.recreate()
# get start/stop nodes
start_nodes, stop_nodes = sgraph.get_start_stop_nodes()
# convert to node intervals
start_nodes = set(sgraph.get_node_interval(n_id) for n_id in start_nodes)
stop_nodes = set(sgraph.get_node_interval(n_id) for n_id in stop_nodes)
assert Exon(959214, 959316) in start_nodes
assert Exon(959316, 964528) in start_nodes
assert Exon(957273, 957434) in start_nodes
assert Exon(944278, 944321) in stop_nodes
# ensure best path uses change points
config = Config.defaults()
config.max_paths = 1
gene_isoforms = assemble_isoforms(sgraph, config)
assert len(gene_isoforms) == 1
isoforms = gene_isoforms[0]
assert len(isoforms) == 1
isoform = isoforms[0]
assert isoform.path[0] == Exon(944321, 944800)
assert isoform.path[-1] == Exon(959214, 959316)
def test_ccle55_cuff_noc2l():
'''Locus containing from 55 CCLE samples assembled with Cufflinks'''
# pull SpliceGraph out of GTF
t_dict, locus = read_single_locus('noc2l_locus.gtf')
found_sgraph = False
for sgraph in locus.create_splice_graphs():
if (sgraph.chrom == 'chr1' and sgraph.start == 934942
and sgraph.end == 976702 and sgraph.strand == Strand.NEG):
found_sgraph = True
break
assert found_sgraph
# examine specific change points
trim = False
pval = 0.1
fc_cutoff = 0.8
n1 = Exon(934942, 944589)
n1_id = sgraph.get_node_id(n1)
assert sgraph.G.is_stop[n1_id]
cps = sgraph.detect_change_points(pval=pval, fc_cutoff=fc_cutoff)
for cp in cps:
sgraph.apply_change_point(cp, trim=trim)
true_starts = set([964528, 957434, 959316])
true_stops = set([944278])
assert true_starts.issubset(sgraph.start_sites)
assert true_stops.issubset(sgraph.stop_sites)
# rebuild graph and examine start/stop nodes
sgraph.recreate()
# get start/stop nodes
start_nodes, stop_nodes = sgraph.get_start_stop_nodes()
# convert to node intervals
start_nodes = set(sgraph.get_node_interval(n_id) for n_id in start_nodes)
stop_nodes = set(sgraph.get_node_interval(n_id) for n_id in stop_nodes)
assert Exon(959214, 959316) in start_nodes
assert Exon(959316, 964528) in start_nodes
assert Exon(957273, 957434) in start_nodes
assert Exon(944278, 944321) in stop_nodes
# ensure best path uses change points
pgf = PathGraphFactory(sgraph)
pgraph, k = pgf.create_optimal()
paths = find_paths(pgraph, max_paths=1)
assert len(paths) == 1
path, expr = paths[0]
path = reconstruct_path(path, pgraph, sgraph)
assert path[0] == Exon(944321, 944800)
assert path[-1] == Exon(959214, 959316)
def test_path_graph1():
# read transcripts
t_dict, locus = read_single_locus('path1.gtf')
SG = SpliceGraph.create(t_dict.values())
# paths
ABCDE = (SOURCE, Exon(0, 100), Exon(200, 300), Exon(400, 500),
Exon(600, 700), Exon(800, 900), SINK)
ACE = (SOURCE, Exon(0, 100), Exon(400, 500), Exon(800, 900), SINK)
ABCE = (SOURCE, Exon(0, 100), Exon(200, 300), Exon(400,
500), Exon(800,
900), SINK)
ACDE = (SOURCE, Exon(0, 100), Exon(400, 500), Exon(600,
700), Exon(800,
900), SINK)
paths = [ABCDE, ACE, ABCE, ACDE]
# create path graph k = 2
k = 2
G1 = create_path_graph(SG, k)
G2 = nx.DiGraph()
for path in paths:
kmers = list(get_kmers(path, k))
add_path(G2, kmers, 1.0)
assert nx.is_isomorphic(G1, G2)
def test_mark_start_stop_sites1():
t_dict, locus = read_single_locus('change_point1.gtf')
sgraph = SpliceGraph.create(t_dict.values())
G = sgraph.G
assert len(G) == 1
n_id = sgraph.get_node_id(Exon(50, 200))
assert sgraph.G.is_start[n_id]
assert sgraph.G.is_stop[n_id]
# add a start site change point
sgraph.start_sites.add(125)
sgraph.recreate()
G = sgraph.G
assert len(G) == 2
n_id = sgraph.get_node_id(Exon(50, 125))
assert sgraph.G.is_start[n_id]
assert not sgraph.G.is_stop[n_id]
n_id = sgraph.get_node_id(Exon(125, 200))
assert sgraph.G.is_start[n_id]
assert sgraph.G.is_stop[n_id]
# add a stop site change point
sgraph.stop_sites.add(80)
sgraph.recreate()
G = sgraph.G
assert len(G) == 3
n_id = sgraph.get_node_id(Exon(50, 80))
assert sgraph.G.is_start[n_id]
assert sgraph.G.is_stop[n_id]
n_id = sgraph.get_node_id(Exon(80, 125))
assert not sgraph.G.is_start[n_id]
assert not sgraph.G.is_stop[n_id]
n_id = sgraph.get_node_id(Exon(125, 200))
assert sgraph.G.is_start[n_id]
assert sgraph.G.is_stop[n_id]
# flip strand
for t_id, t in t_dict.iteritems():
t.strand = Strand.NEG
sgraph = SpliceGraph.create(t_dict.values())
G = sgraph.G
assert len(G) == 1
n_id = sgraph.get_node_id(Exon(50, 200))
assert sgraph.G.is_start[n_id]
assert sgraph.G.is_stop[n_id]
# add a start site change point
sgraph.start_sites.add(125)
sgraph.recreate()
G = sgraph.G
assert len(G) == 2
n_id = sgraph.get_node_id(Exon(50, 125))
assert sgraph.G.is_start[n_id]
assert sgraph.G.is_stop[n_id]
n_id = sgraph.get_node_id(Exon(125, 200))
assert sgraph.G.is_start[n_id]
assert not sgraph.G.is_stop[n_id]
# add a stop site change point
sgraph.stop_sites.add(80)
sgraph.recreate()
G = sgraph.G
assert len(G) == 3
n_id = sgraph.get_node_id(Exon(50, 80))
assert not sgraph.G.is_start[n_id]
assert sgraph.G.is_stop[n_id]
n_id = sgraph.get_node_id(Exon(80, 125))
assert sgraph.G.is_start[n_id]
assert sgraph.G.is_stop[n_id]
n_id = sgraph.get_node_id(Exon(125, 200))
assert sgraph.G.is_start[n_id]
assert not sgraph.G.is_stop[n_id]
def test_mark_start_stop_sites2():
# pos strand not guided
t_dict, locus = read_single_locus('multi_strand1.gtf')
sgraph = SpliceGraph.create(locus.get_transfrags(Strand.POS))
G = sgraph.G
assert G.is_start[sgraph.get_node_id(Exon(100, 200))]
assert G.is_stop[sgraph.get_node_id(Exon(400, 650))]
# neg strand not guided
sgraph = SpliceGraph.create(locus.get_transfrags(Strand.NEG))
G = sgraph.G
assert G.is_start[sgraph.get_node_id(Exon(950, 980))]
assert G.is_stop[sgraph.get_node_id(Exon(400, 500))]
# pos strand guided
sgraph = SpliceGraph.create(locus.get_transfrags(Strand.POS),
guided_ends=True,
guided_assembly=True)
G = sgraph.G
assert G.is_start[sgraph.get_node_id(Exon(100, 150))]
assert G.is_start[sgraph.get_node_id(Exon(150, 200))]
assert G.is_stop[sgraph.get_node_id(Exon(500, 600))]
assert G.is_stop[sgraph.get_node_id(Exon(600, 650))]
assert G.is_ref[sgraph.get_node_id(Exon(150, 200))]
assert G.is_ref[sgraph.get_node_id(Exon(300, 400))]
assert G.is_ref[sgraph.get_node_id(Exon(500, 600))]
assert not G.is_ref[sgraph.get_node_id(Exon(100, 150))]
assert not G.is_ref[sgraph.get_node_id(Exon(600, 650))]
# neg strand guided
sgraph = SpliceGraph.create(locus.get_transfrags(Strand.NEG),
guided_ends=True,
guided_assembly=True)
G = sgraph.G
assert G.is_stop[sgraph.get_node_id(Exon(350, 400))]
assert G.is_start[sgraph.get_node_id(Exon(980, 1000))]
assert not G.is_start[sgraph.get_node_id(Exon(950, 980))]
for n_id in G.node_ids_iter():
assert G.is_ref[n_id]
def test_mark_start_stop_sites2():
# pos strand not guided
t_dict, locus = read_single_locus('multi_strand1.gtf')
sgraph = SpliceGraph.create(locus.get_transfrags(Strand.POS))
G = sgraph.G
assert G.node[sgraph.get_node_id(Exon(100, 200))][SGNode.IS_START]
assert G.node[sgraph.get_node_id(Exon(400, 650))][SGNode.IS_STOP]
# neg strand not guided
sgraph = SpliceGraph.create(locus.get_transfrags(Strand.NEG))
G = sgraph.G
assert G.node[sgraph.get_node_id(Exon(950, 980))][SGNode.IS_START]
assert G.node[sgraph.get_node_id(Exon(400, 500))][SGNode.IS_STOP]
# pos strand guided
sgraph = SpliceGraph.create(locus.get_transfrags(Strand.POS),
guided_ends=True,
guided_assembly=True)
G = sgraph.G
assert G.node[sgraph.get_node_id(Exon(100, 150))][SGNode.IS_START]
assert G.node[sgraph.get_node_id(Exon(150, 200))][SGNode.IS_START]
assert G.node[sgraph.get_node_id(Exon(500, 600))][SGNode.IS_STOP]
assert G.node[sgraph.get_node_id(Exon(600, 650))][SGNode.IS_STOP]
assert G.node[sgraph.get_node_id(Exon(150, 200))][SGNode.IS_REF]
assert G.node[sgraph.get_node_id(Exon(300, 400))][SGNode.IS_REF]
assert G.node[sgraph.get_node_id(Exon(500, 600))][SGNode.IS_REF]
assert not G.node[sgraph.get_node_id(Exon(100, 150))][SGNode.IS_REF]
assert not G.node[sgraph.get_node_id(Exon(600, 650))][SGNode.IS_REF]
# neg strand guided
sgraph = SpliceGraph.create(locus.get_transfrags(Strand.NEG),
guided_ends=True,
guided_assembly=True)
G = sgraph.G
assert G.node[sgraph.get_node_id(Exon(350, 400))][SGNode.IS_STOP]
assert G.node[sgraph.get_node_id(Exon(980, 1000))][SGNode.IS_START]
assert not G.node[sgraph.get_node_id(Exon(950, 980))][SGNode.IS_START]
for n, nd in G.nodes_iter(data=True):
assert nd[SGNode.IS_REF]
return