def sort_by_tree(self):
tree_filename = tkFileDialog.askopenfilename(initialdir = self.host.settings.work_dir, filetypes = (("Inkscape vector file", "*.svg"), ))
if tree_filename == "": # Cancel
return
old_order_seqs = Aln_basic.read_fasta_from_strings(self.fixed.get_strings())
sort_and_color_path = os.path.join(self.host.settings.script_dir, "sort_and_color.py")
fixed_filename = os.path.join(self.host.settings.work_dir, "%s.fixed" % self.host.temp_name)
Aln_basic.write_widget_into_file(self.fixed.text_widget, fixed_filename)
if self.host.verbose.get():
os.system("%s -i %s -w %s -o %s -t %s" % (sort_and_color_path, os.path.basename(fixed_filename), self.host.settings.work_dir,
self.host.temp_name, tree_filename))
else:
os.system("%s -i %s -w %s -o %s -t %s 1> nul 2> nul" % (sort_and_color_path, os.path.basename(fixed_filename),
self.host.settings.work_dir, self.host.temp_name, tree_filename))
temp_sorted_filename = os.path.join(self.host.settings.work_dir, "%s.tree_sorted" % self.host.temp_name)
Aln_basic.read_widget_from_file(self.fixed.text_widget, temp_sorted_filename)
self.IDs.text_widget.delete(1.0, tkinter.END)
new_order_seqs = Aln_basic.read_fasta_from_strings(self.fixed.get_strings())
new_ids = ""
for s in new_order_seqs:
new_ids += "%s\n" % s.ID
self.IDs.text_widget.insert(tkinter.END, new_ids)
print (" Alignment was sorted according to the %s tree file!" % tree_filename)
self.host.log_tab.write_to_log("Alignment was sorted according to the tree file:\n%s" % tree_filename, True)
if len(new_order_seqs) != len(old_order_seqs):
print (" Number of sequences reduced from %i to %i! Possibly IDs in the tree file was interpreted badly" % (len(old_order_seqs), len(new_order_seqs)))
self.host.set_status("Number of sequences reduced from %i to %i!" % (len(old_order_seqs), len(new_order_seqs)))
def obtain_features(self):
print(" Obtaining features...")
curr_mode = self.feature_mode.get()
domain_filename = os.path.join(self.host.settings.work_dir,
"%s.domain_table" % self.host.temp_name)
domain_not_empty = None
if curr_mode == "COG":
domain_not_empty = Aln_basic.write_widget_into_file(
self.hmmresults_COG.text_widget, domain_filename)
else:
domain_not_empty = Aln_basic.write_widget_into_file(
self.hmmresults_Pfam.text_widget, domain_filename)
fixed_filename = os.path.join(self.host.settings.work_dir,
"%s.fixed" % self.host.temp_name)
dom_filename = "%s.domain_info" % self.host.temp_name
Aln_basic.write_widget_into_file(self.host.parse_tab.fixed.text_widget,
fixed_filename)
obtain_features_path = os.path.join(self.host.settings.script_dir,
"obtain_features.py")
data_for_features = ""
if domain_not_empty:
data_for_features += "-p %s" % os.path.basename(domain_filename)
if not self.TMHMM_results.text_is_empty():
TM_filename = os.path.join(self.host.settings.work_dir,
"%s.TMHMM" % self.host.temp_name)
Aln_basic.write_widget_into_file(self.TMHMM_results.text_widget,
TM_filename)
data_for_features += " -t %s" % os.path.basename(TM_filename)
self.host.set_status("Working")
if self.host.verbose.get():
os.system("%s -i %s -w %s -o %s %s -e %s -f %s -d %s" %
(obtain_features_path, os.path.basename(fixed_filename),
self.host.settings.work_dir,
"%s.features" % self.host.temp_name, data_for_features,
self.evalue_threshold.get(),
self.overlap_threshold.get(), dom_filename))
else:
os.system(
"%s -i %s -w %s -o %s %s -e %s -f %s -d %s 1> nul 2> nul" %
(obtain_features_path, os.path.basename(fixed_filename),
self.host.settings.work_dir,
"%s.features" % self.host.temp_name, data_for_features,
self.evalue_threshold.get(), self.overlap_threshold.get(),
dom_filename))
self.host.set_status("Ready")
Aln_basic.read_widget_from_file(
self.features.text_widget,
os.path.join(self.host.settings.work_dir,
"%s.features" % self.host.temp_name))
dom_filename_full = os.path.join(self.host.settings.work_dir,
dom_filename)
if os.path.isfile(
dom_filename_full): # File was created (non-empty domains)
domain_dict = Aln_basic.read_domain_info_file(dom_filename_full)
self.host.load_domain_info(
domain_dict) # Loading domain info into the info tab
print(" [..DONE..]")
def align(self):
if self.seq_input_frame.text_is_empty(): # No sequences were provided
self.host.set_status("No sequences were provided to align!", "#FF0000")
return
print (" Alignment construction started...")
(muscle_name, muscle_path) = Settings.get_program_name(self.host.settings.muscle_dir, "muscle")
unaligned_filename = os.path.join(self.host.settings.work_dir, "%s.fasta" % self.host.temp_name)
Aln_basic.write_widget_into_file(self.seq_input_frame.text_widget, unaligned_filename)
aligned_filename = os.path.join(self.host.settings.work_dir, "%s.aln" % self.host.temp_name)
self.host.set_status("Alignment")
maxiters_option = ""
if self.maxiters.get() != "":
try:
maxiters_option = "-maxiters %i" % int(self.maxiters.get())
except TypeError:
print ("Option -maxiters is not an integer and is ignored!")
gapopen_option = ""
if self.gapopen.get() != "":
if Aln_basic.is_negative_float(self.gapopen.get(), "-gapopen"):
gapopen_option = "-gapopen %s" % self.gapopen.get()
gapextend_option = ""
if self.gapextend.get() != "":
if Aln_basic.is_negative_float(self.gapextend.get(), "-gapextend"):
gapextend_option = "-gapextend %s" % self.gapextend.get()
muscle_command = "%s -in %s -out %s %s %s %s" % (muscle_path, unaligned_filename, aligned_filename, maxiters_option, gapopen_option, gapextend_option)
print ("Muscle command to be ran:")
print (muscle_command)
if self.host.verbose.get():
os.system(muscle_command)
else:
os.system("%s 1> nul 2> nul" % muscle_command)
Aln_basic.read_widget_from_file(self.aln_input_frame.text_widget, aligned_filename)
if self.insert_blocks.get(): # Empty sequence >BLOCKS should be added
curr_seqs = Aln_basic.read_fasta_from_strings(self.aln_input_frame.get_strings())
self.aln_input_frame.text_widget.delete(1.0, tkinter.END)
upd_aln_file = open(aligned_filename, "w")
upd_aln_file.write(">BLOCKS\n")
upd_aln_file.write("%s\n\n" % ("-" * len(curr_seqs[0].sequence)))
upd_aln_file.write(">SITE\n")
upd_aln_file.write("%s\n\n" % ("-" * len(curr_seqs[0].sequence)))
for s in curr_seqs:
s.print_fasta(upd_aln_file, 60)
upd_aln_file.close()
Aln_basic.read_widget_from_file(self.aln_input_frame.text_widget, aligned_filename)
self.host.set_status("Ready")
os.remove(unaligned_filename)
os.remove(aligned_filename)
print (" [..DONE..]")
def check_pending(self):
files_ready = list()
i = 0
while i < len(self.pending_filenames):
filename = self.pending_filenames[i]
if os.path.isfile(filename):
if os.path.getctime(filename) != os.path.getmtime(filename):
files_ready.append(filename)
self.pending_filenames.pop(i)
i -= 1
i += 1
if len(self.pending_filenames) == 0:
self.disable_check_button()
ready_and_required = list()
for curr_file in files_ready:
base_name = os.path.basename(curr_file)
name_parts = base_name.split(".")
if len(
name_parts
) != 2: # Files without extension or with multiple dots are not considered
continue
project_name = name_parts[0]
extension = name_parts[1]
if project_name == self.get_project_name():
if extension == "Pfam_table":
Aln_basic.read_widget_from_file(
self.features_tab.hmmresults_Pfam.text_widget,
curr_file)
ready_and_required.append(extension)
if extension == "COG_table":
Aln_basic.read_widget_from_file(
self.features_tab.hmmresults_COG.text_widget,
curr_file)
ready_and_required.append(extension)
if extension == "TMHMM":
Aln_basic.read_widget_from_file(
self.features_tab.TMHMM_results.text_widget, curr_file)
ready_and_required.append(extension)
if len(ready_and_required) == 0: # No files are ready
self.set_status("Sorry, no file delivered!")
else:
self.set_status("These files were ready and loaded: %s" %
",".join(ready_and_required))
def load_project(self):
project_dir = os.path.join(self.settings.work_dir,
self.get_project_name())
if self.get_project_name() == "": # No project name was entered
project_dir = tkFileDialog.askdirectory(
initialdir=self.settings.work_dir,
title="Please select a folder with the project:")
if project_dir == "": # Cancel
return
if not os.path.isdir(project_dir):
self.set_status(
"This project does not exist, please check its name again!",
"#FF0000")
return
#answer = tkMessageBox.askyesno("Please confirm loading", "Are you sure you want to load %s project? All existing data in the frames, if any, will be re-writed!" % self.get_project_name())
#if answer != True:
# return
self.clear_all()
self.project_title_widget.insert(tkinter.END,
os.path.basename(project_dir))
self.set_status("Working")
print("-------- Project %s is now loading! --------" %
self.get_project_name())
project_files = os.listdir(project_dir)
extension_to_widget = {
"aln": self.input_tab.aln_input_frame.text_widget,
"sample": self.input_tab.seq_input_frame.text_widget,
"fixed": self.parse_tab.fixed.text_widget,
"pure": self.parse_tab.pure.text_widget,
"blocks_regions": self.parse_tab.blocks.text_widget,
"ids": self.parse_tab.IDs.text_widget,
"COG_table": self.features_tab.hmmresults_COG.text_widget,
"Pfam_table": self.features_tab.hmmresults_Pfam.text_widget,
"TMHMM": self.features_tab.TMHMM_results.text_widget,
#"features" : self.features_tab.features.text_widget, #FIX: version 0.2.8 (features should not be loaded)
"auto_log": self.log_tab.auto_log.text_widget,
"man_log": self.log_tab.manual_log.text_widget
}
for curr_file in project_files:
name_parts = curr_file.split(".")
if len(
name_parts
) != 2: # Files without extension or with multiple dots are not considered
continue
project_name = name_parts[0]
extension = name_parts[1]
full_filename = os.path.join(project_dir, curr_file)
if project_name == self.get_project_name():
if extension in extension_to_widget:
Aln_basic.read_widget_from_file(
extension_to_widget[extension], full_filename)
if extension == "pure":
self.parse_tab.enable_pure_analysis()
if extension == "actions":
self.purify_tab.load_actions(full_filename)
if extension == "colors":
self.load_colors_from_file(full_filename)
self.parse_tab.check_numbers()
self.log_tab.color_important()
self.set_status("Ready")
def parse(self):
aligned_filename = os.path.join(self.settings.work_dir,
"%s.aln" % self.temp_name)
Aln_basic.write_widget_into_file(
self.input_tab.aln_input_frame.text_widget, aligned_filename)
remove_seq_limits_path = os.path.join(self.settings.script_dir,
"remove_seq_limits.py")
self.set_status("Working")
if self.verbose.get():
os.system("%s -i %s -w %s -o %s -d -x" %
(remove_seq_limits_path, "%s.aln" % self.temp_name,
self.settings.work_dir, self.temp_name))
else:
os.system("%s -i %s -w %s -o %s -d -x 1> nul 2> nul" %
(remove_seq_limits_path, "%s.aln" % self.temp_name,
self.settings.work_dir, self.temp_name))
self.set_status("Ready")
try:
os.remove(
os.path.join(self.settings.work_dir,
"%s.aln" % self.temp_name))
os.remove(
os.path.join(self.settings.work_dir,
"%s.blocks" % self.temp_name))
os.remove(
os.path.join(self.settings.work_dir,
"%s.motif" % self.temp_name))
os.remove(
os.path.join(self.settings.work_dir,
"%s.motif_var" % self.temp_name))
os.remove(
os.path.join(self.settings.work_dir,
"%s.orgs" % self.temp_name))
os.remove(
os.path.join(self.settings.work_dir,
"%s.pure.correspond" % self.temp_name))
except OSError:
pass
fixed_filename = os.path.join(self.settings.work_dir,
"%s.fixed" % self.temp_name)
Aln_basic.read_widget_from_file(self.parse_tab.fixed.text_widget,
fixed_filename)
os.remove(fixed_filename)
pure_filename = os.path.join(self.settings.work_dir,
"%s.pure" % self.temp_name)
Aln_basic.read_widget_from_file(self.parse_tab.pure.text_widget,
pure_filename)
self.parse_tab.enable_pure_analysis()
os.remove(pure_filename)
blocks_filename = os.path.join(self.settings.work_dir,
"%s.blocks_regions" % self.temp_name)
if os.path.isfile(blocks_filename):
Aln_basic.read_widget_from_file(self.parse_tab.blocks.text_widget,
blocks_filename)
os.remove(blocks_filename)
ids_filename = os.path.join(self.settings.work_dir,
"%s.ids" % self.temp_name)
Aln_basic.read_widget_from_file(self.parse_tab.IDs.text_widget,
ids_filename)
os.remove(ids_filename)
self.parse_tab.check_numbers()