def test_bedfile_to_multivec():
runner = clt.CliRunner()
input_file = op.join(testdir, "sample_data", "sample.multival.bed")
chromsizes_file = op.join(testdir, "sample_data",
"sample.multival.chrom.sizes")
with tempfile.TemporaryDirectory() as tmp_dir:
out_file = op.join(tmp_dir, "out.multivec")
_ = runner.invoke(
ccc.bedfile_to_multivec,
[
input_file,
"--output-file",
out_file,
"--assembly",
"hg38",
"--num-rows",
3,
"--chromsizes-filename",
chromsizes_file,
"--starting-resolution",
"1000",
],
)
# import traceback
# a, b, tb = result.exc_info
# print("exc_info:", result.exc_info)
# print("result:", result)
# print("result.output", result.output)
# print("result.error", traceback.print_tb(tb))
# print("Exception:", a,b)
tsinfo = ctv.tileset_info(out_file)
# print("tsinfo:", tsinfo)
assert "resolutions" in tsinfo
assert tsinfo["max_pos"][0] == 18000
tile = ctv.get_single_tile(out_file, (0, 0))
# input_file:
# chr1 0 1000 1.0 2.0 3.0
# chr1 1000 2000
# chr2 5000 6000 20.0 30.0 40.0
#
# # input chromsizes
# chr1 10000
# chr2 8000
# first row, first chrom first value
assert len(tile) == 3
assert tile[0][0] == 1.0
assert tile[0][15] == 20.0
assert tile[1][0] == 2.0
assert tile[2][0] == 3.0
assert tile[2][15] == 40.0
def test_multivec():
filename = op.join("test/sample_data", "sample_gwas.multires.mv5")
with h5py.File(filename, "r") as h5:
tile_size = h5["info"].attrs["tile-size"]
resolutions = list(h5["resolutions"].keys())
reso = resolutions[0]
chroms = h5[f"resolutions/{reso}/chroms/name"][:]
num_rows = h5[f"resolutions/{reso}/values"][chroms[0]].shape[1]
total_length = sum(h5["chroms/length"])
# info
info = hgmu.tileset_info(filename)
assert info["shape"] == [tile_size, num_rows]
assert info["tile_size"] == tile_size
assert info["max_pos"] == total_length
assert set(info["resolutions"]) == set(int(reso) for reso in resolutions)
# get_single_tile
test_tile = hgmu.get_single_tile(filename, [0, 0])
assert list(test_tile.shape)[::-1] == info["shape"]
with pytest.raises(IndexError):
hgmu.get_single_tile(filename, [len(resolutions), 0])
# tiles
tids = [
f"test_uuid.{level}.0.1231.123" for level in range(len(resolutions))
]
tiles = hgmu.tiles(filename, tids)
for tile_id, tile_value in tiles:
tile_pos = [int(i) for i in tile_id.split(".")[1:3]]
single_tile = hgmu.get_single_tile(filename, tile_pos).astype(
tile_value["dtype"])
assert (base64.b64encode(
single_tile.ravel()).decode("utf-8") == tile_value["dense"])
def tileset_info(request):
"""Get information about a tileset
Tilesets have information critical to their display
such as the maximum number of dimensions and well as
their width. This needs to be relayed to the client
in order for it to know which tiles to request.
Args:
request (django.http.HTTPRequest): The request object
containing tileset_ids in the 'd' parameter.
Return:
django.http.JsonResponse: A JSON object containing
the tileset meta-information
"""
queryset = tm.Tileset.objects.all()
tileset_uuids = request.GET.getlist("d")
tileset_infos = {}
chromsizes_error = None
if "cs" in request.GET:
# we need to call a different server to get the tiles
if not "ci" in request.GET.getlist:
chromsizes_error = "cs param present without ci"
# call the request server and get the chromsizes
pass
else:
if "ci" in request.GET:
try:
chromsizes = tm.Tileset.objects.get(uuid=request.GET["ci"])
data = tcs.chromsizes_array_to_series(
tcs.get_tsv_chromsizes(chromsizes.datafile.path)
)
except Exception as ex:
pass
for tileset_uuid in tileset_uuids:
tileset_object = queryset.filter(uuid=tileset_uuid).first()
if tileset_uuid == "osm-image":
tileset_infos[tileset_uuid] = {
"min_x": 0,
"max_height": 134217728,
"min_y": 0,
"max_y": 134217728,
"max_zoom": 19,
"tile_size": 256,
}
continue
if tileset_object is None:
tileset_infos[tileset_uuid] = {
"error": "No such tileset with uid: {}".format(tileset_uuid)
}
continue
if tileset_object.private and request.user != tileset_object.owner:
# dataset is not public
tileset_infos[tileset_uuid] = {"error": "Forbidden"}
continue
if tileset_object.requiresAuthentication and not request.user.is_authenticated:
# dataset is not public
tileset_infos[tileset_uuid] = {
"error": "This request required authentication"
}
continue
#print(tileset_uuid,"accessibleTilesets" in request.session)
if tileset_object.requiresAuthentication and (
("accessibleTilesets" not in request.session)
or (tileset_uuid not in request.session["accessibleTilesets"])
):
# dataset is not accessible for this user
tileset_infos[tileset_uuid] = {
"error": "You don't have access to this tileset"
}
continue
if tileset_object.filetype == "hitile" or tileset_object.filetype == "hibed":
tileset_info = hdft.get_tileset_info(
h5py.File(tileset_object.datafile.path, "r")
)
tileset_infos[tileset_uuid] = {
"min_pos": [int(tileset_info["min_pos"])],
"max_pos": [int(tileset_info["max_pos"])],
"max_width": 2
** math.ceil(
math.log(tileset_info["max_pos"] - tileset_info["min_pos"])
/ math.log(2)
),
"tile_size": int(tileset_info["tile_size"]),
"max_zoom": int(tileset_info["max_zoom"]),
}
elif tileset_object.filetype == "bigwig":
chromsizes = tgt.get_chromsizes(tileset_object)
tsinfo = hgbi.tileset_info(tileset_object.datafile.path, chromsizes)
# print('tsinfo:', tsinfo)
if "chromsizes" in tsinfo:
tsinfo["chromsizes"] = [(c, int(s)) for c, s in tsinfo["chromsizes"]]
tileset_infos[tileset_uuid] = tsinfo
elif tileset_object.filetype == "bigbed":
chromsizes = tgt.get_chromsizes(tileset_object)
tsinfo = hgbi.tileset_info(tileset_object.datafile.path, chromsizes)
# print('tsinfo:', tsinfo)
if "chromsizes" in tsinfo:
tsinfo["chromsizes"] = [(c, int(s)) for c, s in tsinfo["chromsizes"]]
tileset_infos[tileset_uuid] = tsinfo
elif tileset_object.filetype == "multivec":
tileset_infos[tileset_uuid] = hgmu.tileset_info(
tileset_object.datafile.path
)
elif tileset_object.filetype == "elastic_search":
response = urllib.urlopen(tileset_object.datafile + "/tileset_info")
tileset_infos[tileset_uuid] = json.loads(response.read())
elif tileset_object.filetype == "beddb":
tileset_infos[tileset_uuid] = cdt.get_tileset_info(
tileset_object.datafile.path
)
elif tileset_object.filetype == "bed2ddb":
tileset_infos[tileset_uuid] = cdt.get_2d_tileset_info(
tileset_object.datafile.path
)
elif tileset_object.filetype == "cooler":
tileset_infos[tileset_uuid] = hgco.tileset_info(
tileset_object.datafile.path
)
elif tileset_object.filetype == "time-interval-json":
tileset_infos[tileset_uuid] = hgti.tileset_info(
tileset_object.datafile.path
)
elif (
tileset_object.filetype == "2dannodb"
or tileset_object.filetype == "imtiles"
):
tileset_infos[tileset_uuid] = hgim.get_tileset_info(
tileset_object.datafile.path
)
elif tileset_object.filetype == "geodb":
tileset_infos[tileset_uuid] = hggo.tileset_info(
tileset_object.datafile.path
)
elif tileset_object.filetype == "bam":
tileset_infos[tileset_uuid] = ctb.tileset_info(tileset_object.datafile.path)
tileset_infos[tileset_uuid]["max_tile_width"] = hss.MAX_BAM_TILE_WIDTH
else:
# Unknown filetype
tileset_infos[tileset_uuid] = {
"error": "Unknown filetype " + tileset_object.filetype
}
tileset_infos[tileset_uuid]["name"] = tileset_object.name
tileset_infos[tileset_uuid]["datatype"] = tileset_object.datatype
tileset_infos[tileset_uuid]["coordSystem"] = tileset_object.coordSystem
tileset_infos[tileset_uuid]["coordSystem2"] = tileset_object.coordSystem2
return JsonResponse(tileset_infos)
def tileset_info(request):
''' Get information about a tileset
Tilesets have information critical to their display
such as the maximum number of dimensions and well as
their width. This needs to be relayed to the client
in order for it to know which tiles to request.
Args:
request (django.http.HTTPRequest): The request object
containing tileset_ids in the 'd' parameter.
Return:
django.http.JsonResponse: A JSON object containing
the tileset meta-information
'''
queryset = tm.Tileset.objects.all()
tileset_uuids = request.GET.getlist("d")
tileset_infos = {}
chromsizes_error = None
if 'cs' in request.GET:
# we need to call a different server to get the tiles
if not 'ci' in request.GET.getlist:
chromsizes_error = 'cs param present without ci'
# call the request server and get the chromsizes
pass
else:
if 'ci' in request.GET:
try:
chromsizes = tm.Tileset.objects.get(uuid=request.GET['ci'])
data = tcs.chromsizes_array_to_series(
tcs.get_tsv_chromsizes(chromsizes.datafile.path))
except Exception as ex:
pass
for tileset_uuid in tileset_uuids:
tileset_object = queryset.filter(uuid=tileset_uuid).first()
if tileset_uuid == 'osm-image':
tileset_infos[tileset_uuid] = {
'min_x': 0,
'max_height': 134217728,
'min_y': 0,
'max_y': 134217728,
'max_zoom': 19,
'tile_size': 256
}
continue
if tileset_object is None:
tileset_infos[tileset_uuid] = {
'error': 'No such tileset with uid: {}'.format(tileset_uuid)
}
continue
if tileset_object.private and request.user != tileset_object.owner:
# dataset is not public
tileset_infos[tileset_uuid] = {'error': "Forbidden"}
continue
if (
tileset_object.filetype == 'hitile' or
tileset_object.filetype == 'hibed'
):
tileset_info = hdft.get_tileset_info(
h5py.File(tileset_object.datafile.path, 'r'))
tileset_infos[tileset_uuid] = {
"min_pos": [int(tileset_info['min_pos'])],
"max_pos": [int(tileset_info['max_pos'])],
"max_width": 2 ** math.ceil(
math.log(
tileset_info['max_pos'] - tileset_info['min_pos']
) / math.log(2)
),
"tile_size": int(tileset_info['tile_size']),
"max_zoom": int(tileset_info['max_zoom'])
}
elif tileset_object.filetype == 'bigwig':
chromsizes = tgt.get_chromsizes(tileset_object)
tsinfo = hgbi.tileset_info(
tileset_object.datafile.path,
chromsizes
)
#print('tsinfo:', tsinfo)
if 'chromsizes' in tsinfo:
tsinfo['chromsizes'] = [(c, int(s)) for c,s in tsinfo['chromsizes']]
tileset_infos[tileset_uuid] = tsinfo
elif tileset_object.filetype == 'bigbed':
chromsizes = tgt.get_chromsizes(tileset_object)
tsinfo = hgbi.tileset_info(
tileset_object.datafile.path,
chromsizes
)
#print('tsinfo:', tsinfo)
if 'chromsizes' in tsinfo:
tsinfo['chromsizes'] = [(c, int(s)) for c,s in tsinfo['chromsizes']]
tileset_infos[tileset_uuid] = tsinfo
elif tileset_object.filetype == 'multivec':
tileset_infos[tileset_uuid] = hgmu.tileset_info(
tileset_object.datafile.path)
elif tileset_object.filetype == "elastic_search":
response = urllib.urlopen(
tileset_object.datafile + "/tileset_info")
tileset_infos[tileset_uuid] = json.loads(response.read())
elif tileset_object.filetype == 'beddb':
tileset_infos[tileset_uuid] = cdt.get_tileset_info(
tileset_object.datafile.path
)
elif tileset_object.filetype == 'bed2ddb':
tileset_infos[tileset_uuid] = cdt.get_2d_tileset_info(
tileset_object.datafile.path
)
elif tileset_object.filetype == 'cooler':
tileset_infos[tileset_uuid] = hgco.tileset_info(
tileset_object.datafile.path
)
elif tileset_object.filetype == 'time-interval-json':
tileset_infos[tileset_uuid] = hgti.tileset_info(
tileset_object.datafile.path
)
elif (
tileset_object.filetype == '2dannodb' or
tileset_object.filetype == 'imtiles'
):
tileset_infos[tileset_uuid] = hgim.get_tileset_info(
tileset_object.datafile.path
)
elif tileset_object.filetype == 'geodb':
tileset_infos[tileset_uuid] = hggo.tileset_info(
tileset_object.datafile.path
)
elif tileset_object.filetype == 'bam':
tileset_infos[tileset_uuid] = ctb.tileset_info(
tileset_object.datafile.path
)
tileset_infos[tileset_uuid]['max_tile_width'] = hss.MAX_BAM_TILE_WIDTH
else:
# Unknown filetype
tileset_infos[tileset_uuid] = {
'error': 'Unknown filetype ' + tileset_object.filetype
}
tileset_infos[tileset_uuid]['name'] = tileset_object.name
tileset_infos[tileset_uuid]['datatype'] = tileset_object.datatype
tileset_infos[tileset_uuid]['coordSystem'] = tileset_object.coordSystem
tileset_infos[tileset_uuid]['coordSystem2'] =\
tileset_object.coordSystem2
return JsonResponse(tileset_infos)
def test_multivec():
filename = op.join('data', 'all.KL.bed.multires.mv5')
tsinfo = hgmu.tileset_info(filename)
