def generate_output(self):
"""
Majority of Cotranscriptional SHAPE-Seq output is created here. This includes the DG plot, best structure images, and the movie of the best structures.
"""
draw_dir = OSU.create_directory(self.output_dir + "/draw/")
nn_dir = OSU.create_directory(self.output_dir + "/nn/")
OSU.create_directory(nn_dir + "distances/")
sorted_lengths = sorted(self.file_data)
zero_padding = int(math.floor(math.log10(sorted_lengths[-1])) + 1)
# Parallelized function calls to generate DG plot, distance matrices for clustering of structures, and creating images of minimum distance structures
draw_struct_nums = [
length for length in sorted_lengths
if self.file_data[length]["rc_flag"]
]
draw_args_pool = zip([self.file_data[dsn] for dsn in draw_struct_nums],
draw_struct_nums, repeat(sorted_lengths[-1]),
range(1,
len(draw_struct_nums) + 1),
repeat(zero_padding), repeat(draw_dir),
repeat(self.output_dir + "/ct/"),
repeat(self.draw_all),
repeat(self.most_count_tie_break))
args_pool = [
(PCSU.generate_DG_output, (self, 1, sorted_lengths[-1]))
] + zip(repeat(PCSU.generate_best_struct_images), draw_args_pool)
if self.p == 1:
for i in range(len(args_pool)):
PCSU.calculate_function_helper(args_pool[i])
else:
PCSU.generate_DG_output(
self, 1, sorted_lengths[-1]
) # moved this outside of multiprocessing because Quest has issues running it in a pool
PCSU.run_output_multiprocessing_pool(
PCSU.calculate_function_helper, args_pool[1:], self.p)
if not OSU.check_file_exists(
self.output_dir + "/DG_state_plot.pdf"
): # Weird error on quest that it will ignore this command if sample size is very large
PCSU.generate_DG_output(self, 1, sorted_lengths[-1])
# Use ffmpeg on varna_num.png's to create video of the minimum distance folding pathway
OSU.make_symbolic_link(
draw_dir + str(len(draw_struct_nums)).zfill(zero_padding) +
"_structure.png",
draw_dir + str(len(draw_struct_nums) + 1).zfill(zero_padding) +
"_structure.png") # ffmpeg needs a duplicate of the last frame
VIU.generate_movie(draw_dir + "%%%dd_structure.png" % (zero_padding),
self.output_dir + "/movie.mp4")
return
def generate_best_struct_images(cotrans_length, length, longest_length,
varna_num, zero_padding, draw_dir, ct_dir,
draw_all, most_count_tie_break):
"""
Generates images of the minimum distance structures at a given length
"""
print "generate_best_struct_images: " + str(length)
fname = cotrans_length["filename"]
rho_varna = "\"" + ";".join([str(r) for r in cotrans_length["rho"]]) + "\""
seen_snum = []
mult_images = []
if cotrans_length["rc_flag"]: # plot only structures with a good RC
for snum in cotrans_length["min_dist_indices"]:
seen_snum.append(snum)
for sf in range(len(seen_snum)):
draw_outname_pre = "%s/%s_%s_%s" % (draw_dir, fname, seen_snum[sf],
str(varna_num).zfill(zero_padding))
if len(seen_snum) > 1 and draw_all:
draw_outname_pre += "_mult" + str(sf)
mult_images.append(draw_outname_pre + "_structure.png")
elif len(seen_snum) > 1 and not draw_all:
# draw only the structure with the most supporting counts from the Boltzman samples
structs_str = [",".join(s) for s in cotrans_length["structs"]]
if most_count_tie_break:
tie_break = max(
dict((k, v) for k, v in
cotrans_length["sampled_structs_count"].iteritems()
if structs_str.index(
k) in cotrans_length["min_dist_indices"]))
tie_break_i = structs_str.index(tie_break)
else:
tie_break = min([
cotrans_length["free_energies"][sn]
for sn in cotrans_length["min_dist_indices"]
])
tie_break_i = [
cotrans_length["free_energies"][sn]
for sn in cotrans_length["min_dist_indices"]
].index(tie_break)
tie_break_i = cotrans_length["min_dist_indices"][tie_break_i]
if tie_break_i != seen_snum[sf]:
continue
rho_varna = rho_varna[:-1] + ";".join([""] + ["-1"] *
(longest_length - length)) + "\""
SU.run_ct2dot(ct_dir + fname + "_unique.ct", seen_snum[sf],
draw_outname_pre + ".dbn")
# determine length of .'s needed to fill in the whole length
OSU.system_command(
"sed '$s/$/&%s/' %s.dbn > %s%s_temp.dbn " %
("." *
(longest_length - length), draw_outname_pre, draw_dir, varna_num))
VIU.run_VARNA(draw_dir + str(varna_num) + "_temp.dbn",
draw_outname_pre + "_structure.png",
rho_varna) # same fix this as above
if sf == len(seen_snum) - 1 and len(
seen_snum) > 1 and draw_all: # vertical concat mult images
v_outname = re.sub("_mult\d+", "", mult_images[0])
VIU.vertical_image_concat(v_outname, mult_images)
draw_outname_pre = re.findall("(.*)_structure.png$", v_outname)[0]
if sf == len(seen_snum) - 1 or not draw_all:
print draw_dir + str(varna_num).zfill(
zero_padding) + "_structure.png"
print "SYM LINK: " + draw_dir + str(varna_num).zfill(
zero_padding) + "_structure.png"
OSU.make_symbolic_link(
re.sub("_mult\d+", "", draw_outname_pre) + "_structure.png",
draw_dir + str(varna_num).zfill(zero_padding) +
"_structure.png")
VIU.convert_center_resize(
draw_dir + str(varna_num).zfill(zero_padding) +
"_structure.png", "1200x2800")
OSU.remove_file(draw_dir + str(varna_num) + "_temp.dbn")
return draw_dir + str(varna_num).zfill(zero_padding) + "_structure.png"