sbi_project.py

# python modules
import Bio.PDB
#import numpy
import os
import sys
#import gzip
#import re
import collections as col
import copy
import argparse
import tarfile
import re
import ast
import warnings
#import shutil
from Bio import BiopythonWarning

# our modules
from functions import *
from classes import *
from reduce_inputs_func import *
from get_interactions import *
from DOPE_comparison import * #refine and compare
from DOPE_profile import * #plot energy of one model
import utilities

########################################################################################################################
######################################### Parser for command-line options ##############################################
########################################################################################################################

parser = argparse.ArgumentParser(description="Program to reconstruct a macrocomplex, given a set of interacting pairs "
                                             "(prot-prot, prot-RNA).",
                                 formatter_class=argparse.ArgumentDefaultsHelpFormatter)

parser.add_argument('-i', '--input',
                    dest="indir",
                    action="store",
                    default=None,
                    help="This is a directory provided by the user which contains the inputs pairs (compressed format is "
                         "also available .tar.gz). In case you want to rebuild a macrocomplex from PDB, you can introduce "
                         "its pdb file (.pdb).")


parser.add_argument('-o', '--output',
                    dest="outdir",
                    action="store",
                    default=None,
                    help="This is a directory which will be created during the program, structured in other subdirectories.")


parser.add_argument('-v', '--verbose',
                    dest="verbose",
                    action="store_true",
                    default=False,
                    help="This option will allow the user to completely follow the progam.")

parser.add_argument('-cc', '--clash_distance_cutoff',
                    dest="clash_distance_cutoff",
                    action="store",
                    type=float,
                    default=1.2,
                    choices=[0.25, 0.5, 1, 1.2, 1.5],
                    help="Choose a cut-off distance to detect clashes between chains.")

parser.add_argument('-e', '--energy_plot',
                    dest="energy_plot",
                    action="store_true",
                    default=False,
                    help="DOPE profile of the macrocomplex (not including nucleic acid chains) generated by Modeller.")

parser.add_argument('-ref', '--refine',
                    dest="refine",
                    action="store_true",
                    default=False,
                    help="Refines a model previously generated with the program according to the optimization parameters"
                         " defined by Modeller.")

parser.add_argument('-ver', '--version',
                    dest="version",
                    action="store",
                    type=str,
                    default="short",
                    choices=["short", "intensive"],
                    help="Short version only considers the first model the program can build, while Intensive version"
                         "considers all the possible models the program can build depending on the input pair it starts"
                         "the construction of the model.")

options = parser.parse_args()

########################################################################################################################
######################################### Getting and Organizing the inputs pairs ######################################
########################################################################################################################

print("\nSTARTING THE ANALYSIS\n\n")

# Avoid printing BiopythonWarnings
if options.verbose == False:
    warnings.simplefilter('ignore', BiopythonWarning)

# Output directory
outputs_dir = options.outdir

# Reading inputs files from the user directory (considering .tar.gz)
inputs_dir = options.indir
user_dir = options.indir
p_tar_gz = re.compile(".+\.tar\.gz$")
if options.indir:
    if p_tar_gz.match(options.indir):
        inputs_tar = tarfile.open(options.indir)
        inputs_tar.extractall()
        inputs_dir = [x.name for x in inputs_tar][0]
        inputs_files_path = [x.name for x in inputs_tar][1:]
        inputs_files = [os.getcwd() + "/" + x for x in inputs_files_path]

    elif os.path.isdir(options.indir):
        inputs_dir = options.indir
        inputs_files_path = os.listdir(inputs_dir)
        inputs_files = [os.getcwd() + "/" + inputs_dir + "/" + x for x in
                        inputs_files_path]  # this directory is the one provided by the user
    elif options.indir.endswith(".pdb"):
        if options.verbose:
            sys.stderr.write("The program is getting all the possible pair interactions of the %s macrocomplex" % (inputs_dir))
        inputs_dir = get_interactions(options.indir, outputs_dir)
        inputs_files_path = os.listdir(inputs_dir)
        inputs_files = [os.getcwd() + "/" + inputs_dir + x for x in
                        inputs_files_path]
        if options.verbose:
            sys.stderr.write("A directory named 'get_interactions_results' has been created here:\n %s\n"  % (inputs_dir))

# Checking if the input directory is a directory.
try:
    check_input(inputs_dir, user_dir)
except IncorrectInputDir as error:
    print(str(error))

# Creating a directory (reduced_inputs) in the user output directory (outputs_dir), which contains reduced inputs from the
# user inputs (non-redundant interactions).
reduce_inputs_func(inputs_files, outputs_dir, options.verbose)

# Setting the directory where the files in reduced_inputs directory are.
inputs_reduced = outputs_dir + '/reduced_inputs/'
inputs_reduced_files = os.listdir(inputs_reduced)

if options.verbose:
    sys.stderr.write("A directory named 'reduced_inputs' is created here:\n %s\n\n" % (inputs_reduced))

PDBparser = Bio.PDB.PDBParser()

# Creating a list of dictionaries with the pairs of the inputs. The key is the object with the id changed to numbers
# for handeling the limitation of letters (<chain1>) and the value the id from the pdb file (A).
list_of_dic_inputs = []
c = 0
for pair in inputs_reduced_files:
    chain_pair_dic = {}
    pdb_code = pair
    pdb_filename = inputs_reduced + pair  # .pdb
    structure = PDBparser.get_structure(pdb_code, pdb_filename)
    for model in structure:
        for chain in model:
            c += 1
            original_id = copy.copy(chain.id)
            chain.id = c
            chain_pair_dic[chain] = original_id
    list_of_dic_inputs.append(chain_pair_dic)

# List of tuples where the first element is the chain object and the second, its sequence
chain_seq_list = []
for dic in list_of_dic_inputs:
    chain_seq = tuple()
    for key in dic:
        chain_seq = (key, get_seq_from_pdbchain(key))
        chain_seq_list.append(chain_seq)

if options.verbose:
    sys.stderr.write("Making pairwise comparisons between all the inputs to create a dictionary of chains equivalences\n")

# Making pairwise comparisons between all the inputs to create a dictionary of equivalences (chains_eq_dic).
# Creating the alignment_results.txt file in the user output directory (outputs_dir)
chains_eq_dic = {}
alignments_results = open(outputs_dir + "/alignments_results.txt", "w")

cutoff = 0.9
cutoff_list = []
for element in chain_seq_list:
    element_len = len(element[1])
    for element2 in chain_seq_list:
        element2_len = len(element2[1])
        alignments_results.write("This is the alignment between: %s and %s: " % (element, element2))
        alignments = pairwise2.align.globalxx(element[1], element2[1])
        if len(alignments) > 0:
            score_chain = alignments[0][2]
            len_max = max(element_len, element2_len)
            cutoff_chain = score_chain / len_max
            info = (element[0], element2[0], cutoff_chain)
            cutoff_list.append(info)
            if cutoff_chain >= cutoff:
                alignments_results.write("\tThe max score is %d\n" % (alignments[0][2]))
                chains_eq_dic[element[0]] = element2[0]
                break
        else:
            alignments_results.write("\tNO ALIGNMENT\n")
alignments_results.close()

if options.verbose:
    sys.stderr.write("You can find more information about this steps in 'alignments_results.txt' file here:\n %s\n\n" % (alignments_results.name))

if options.verbose:
    sys.stderr.write("Determining the unique chains to be able to decide the stoichiometry later.\n")

# Set that will contain the unique chains of the complex
unique_chains = set(chains_eq_dic.values())

if options.verbose:
    sys.stderr.write("The program has found %d unique chains.\n" % (len(unique_chains)))

# Assignation of a letter to the unique chains to show them to the user and obtain the stoichiometry as a dictionary (number_to_letter)
number_to_letter = {}
for ch in unique_chains:
    number_to_letter[ch] = ""
    for character in utilities.ascii_list:
        if character not in number_to_letter.values():
            number_to_letter[ch] = character
            break

# Multifasta file containing the unique chains. It will be created as unique_chains_fasta.mfa in the user output directory (outputs_dir),
# to allow the user to introduce the stoichiometry of the macrocomplex.
unique_chains_mfa = unique_chains_fasta(number_to_letter, outputs_dir)

if options.verbose:
    sys.stderr.write("A multifasta file named 'unique_chains_fasta.mfa' of the unique chains has been created here: \n %s\n" % (unique_chains_mfa.name))

# Getting the stoichiometry from the user
while True:
    print("Check the unique_chains_fasta.mfa file to decide the stoichiometry of the macrocomplex.\n ")
    print("PLEASE ENTER THE STOICHIOMETRY\nIn this format: {'a': 1, 'b': 1, 'c': 2}\n ")
    try:
        a = input("Please enter the stoichiometry dictionary: ")
        ref_counter_user = ast.literal_eval(a)
    except ValueError:
        print("Something went wrong, try again...\n")
        continue
    except SyntaxError:
        print("Something went wrong, try again...\n")
        continue
    else:
        if isinstance(ref_counter_user, dict) == False:
            print("Something went wrong, try again...\n")
            continue
    break

if options.verbose:
    sys.stderr.write("The stoichiometry dictionary has correctly been introduced.\n")

# Resetting the value (chain id) of the list_of_dic_inputs keys to its equivalence (obtained above by pairwise alignment)
list_of_dic = []
for dic in list_of_dic_inputs:
    dic_pair = {}
    for key, value in dic.items():
        for key2, value2 in chains_eq_dic.items():
            if key == key2:
                for key3, value3 in number_to_letter.items():
                    if value2 == key3:
                        dic_pair[key] = value3
    list_of_dic.append(dic_pair)

# Converting the list (list_of_dic) into a dictionary (allchains)
allchains = {}
for d in list_of_dic:
    for k, v in d.items():
        allchains[k] = v

# Preventing from problems with clashes (avoiding infinite loops)
n_chains = sum(ref_counter_user.values())
n_pairs = len(list_of_dic)

threshold_clashes = n_chains * 2 * n_pairs

########################################################################################################################
############################################### Construction of the models #############################################
########################################################################################################################
print("\n### STARTING THE CONSTRUCTION OF THE MACROCOMPLEX... Please wait.\n")

# Cut-off distance to detect clashes
clash_distance_cutoff = options.clash_distance_cutoff

# Creating a subdirectory (models) in the user output directory (outputs_dir), which will containg all the models
output_models = (outputs_dir + "/models/")
if not os.path.exists(output_models):
    os.makedirs(output_models)

# Creating as many models as inputs are in the reduced_inputs directory, through iterating among the list_of_dic.
n_model = 0
model_number = 0
for input in list_of_dic:
    n_model += 1
    if options.verbose:
        sys.stderr.write("\n# Trying to build the model %d\n" % (n_model))
    n_round = 0
    clash_count = 0

    #  Selecting the pair of chains that will be the initial reference chains to build the model (ref_counter_chains)
    input_idx = list_of_dic.index(input)
    ref_counter_chains = copy.copy(list_of_dic[input_idx])

    # Change ids of the reference chains (from numbers to letters) to save the model (.pdb) with a uniform format.
    input_ids = ["a", "b"]
    i = 0
    for k, v in copy.copy(list_of_dic[input_idx]).items():
        k.id = input_ids[i]
        i += 1

    # Creating the current model with the reference chains
    current_model = Bio.PDB.Model.Model(model_number)
    for chain in copy.copy(list_of_dic[input_idx]).keys():
        current_model.add(chain)

    # Adding the model to the list of models
    list_models = [current_model]

    # Define initial stoichiometry counter from the stoichiometry introduced by the user (ref_counter_user)
    ref_counter = {}
    for ids, count in ref_counter_user.items():
        ref_counter[ids] = count
        i = count
        for key, value in copy.copy(list_of_dic[input_idx]).items():
            if value == ids:
                i -= 1
                ref_counter[ids] = i

    #  Selecting all other input pairs (total_list) except from the reference input pair (ref_chains)
    ref_chains = copy.copy(list_of_dic[input_idx])
    idx = list_of_dic.index(ref_chains)
    total_list = list_of_dic[:idx] + list_of_dic[idx + 1:]

    # Checking the number of chains in ref_chains and in total_list to ensure that the model can be build based on the
    # stoichiometry

    # Counting of ref_chains
    ref_chains_count = {}
    for v in ref_chains.values():
        if not v in ref_chains_count:
            ref_chains_count[v] = 1
        else:
            ref_chains_count[v] += 1

    # Creation and counting of total_dict
    total_dict = {}

    for x in total_list:
        for k, v in x.items():
            total_dict[k] = v

    total_dict_count = {}
    for ch, letter in allchains.items():
        total_dict_count[letter] = 0

    for v in total_dict.values():
        if v not in total_dict_count:
            total_dict_count[v] = 1
        else:
            total_dict_count[v] += 1

    # Checking the actual counts
    check_input = 0
    for k, v in ref_chains.items():
        if ref_counter[v] == 0:
            check_input += 1
        elif ref_counter[v] >= 1:
            if ref_chains_count[v] == 2 or total_dict_count[v] >= 1:
                check_input += 1

    if check_input == 2:

        # Defining a boolean flag (run) to check if there are still remaining chains to be added to the model.
        run = True
        while run == True:

            # Iterating through each dictionary (alt_chains) of the total_list
            for alt_chains in total_list:

                # Iterating through each chain of ref_chains to compare it with the current dictionary (alt_chains)
                for key, value in ref_chains.items():
                    n_round += 1

                    for element in alt_chains:

                        # Checking if the ids from the current compared chains (ref_chains vs. alt_chains) are equal
                        # to proceed with the analysis
                        if value == alt_chains[element]:

                            # Defining variables for the further superimposition (fixedchain, movingchain and altchain)
                            fixedchain = key
                            movingchain = element
                            alt_chains_copy = col.OrderedDict(alt_chains)
                            del alt_chains_copy[movingchain]
                            altchain = list(alt_chains_copy)[0]


                            # Saving actual altchain (i_chain) to lately check it in the counter
                            i_chain = alt_chains[altchain]

                            for ele in ref_counter:
                                # Checking if there are still chains with the same id as i_chain to be added to the model
                                if ref_counter[i_chain] != 0:

                                    # Getting atoms list for the superimposition
                                    fixed_atoms_list = get_atoms_list(fixedchain)
                                    moving_atoms_list = get_atoms_list(movingchain)
                                    alt_atoms_list = get_atoms_list(altchain)

                                    # Creating a list with the used ids (letters) of the current model chains. Using
                                    # this list to control the "same id error" when adding a new chain
                                    used_letters_list = [x.id for x in current_model]

                                    # If the length exceeds the maximum different characters available, a new empty
                                    # model will be created to be able to add chains avoiding this error
                                    if len(used_letters_list) == 83:
                                        list_models.append(current_model)
                                        io = Bio.PDB.PDBIO()
                                        io.set_structure(current_model)
                                        model_number += 1
                                        current_model = Bio.PDB.Model.Model(model_number)
                                        used_letters_list = []

                                    # Creating a copy of the altchain to add it with a new avaiable id
                                    for character in utilities.ascii_list:
                                        if character not in used_letters_list:
                                            new_id = character
                                            break
                                    new_chain = Bio.PDB.Chain.Chain(new_id)
                                    altchain = copy.copy(altchain)
                                    for residue in altchain.get_residues():
                                        new_chain.add(residue.copy())

                                    current_model.add(new_chain)

                                    # Comparing number of atoms from fixed and moving chain. If it is different,
                                    # refining the lists to obtain equal lenght
                                    if len(fixed_atoms_list) != len(moving_atoms_list):
                                        chains_pattern = refine_for_superimpose(fixedchain, movingchain)
                                        fixed_pattern = chains_pattern[0]
                                        moving_pattern = chains_pattern[1]

                                        fixedchain = get_chain_refined(fixedchain, fixed_pattern)
                                        movingchain = get_chain_refined(movingchain, moving_pattern)

                                        fixed_atoms_list = get_atoms_list(fixedchain)
                                        moving_atoms_list = get_atoms_list(movingchain)

                                    # Superimposing the chains
                                    super_imposer = Bio.PDB.Superimposer()
                                    super_imposer.set_atoms(fixed_atoms_list, moving_atoms_list)
                                    super_imposer.apply(new_chain.get_atoms())

                                    # Checking if the added chain clashes with any chain of the current model. If it
                                    # clashes it will be detached
                                    clash = check_clashes_multi(current_model, list_models, new_chain,
                                                                clash_distance_cutoff)

                                    if clash == True:
                                        clash_count += 1
                                        current_model.detach_child(new_chain.id)
                                        break

                                    if options.verbose:
                                        sys.stderr.write("\tA new chain has been added to the model.\n")

                                    # Actualize ref_conter_chains (adding i_chain)
                                    ref_counter_chains[new_chain] = i_chain

                                    # Actualize ref_counter considering the previously added i_chain
                                    for ids, count in ref_counter.items():
                                        if ids == i_chain:
                                            i = count
                                            if i != 0:
                                                i -= 1
                                                ref_counter[i_chain] = i
                                    if options.verbose:
                                        sys.stderr.write("\tThe following chains has not been yet included in the model:\n\t  %s\n " % ref_counter)
                                    break

            if clash_count >= threshold_clashes:
                print("\nSomething went wrong...\n")
                if options.version == "intensive":
                    sys.stderr.write("Trying to reconstruct the model from the following input.\n")
                break

            ref_chains = copy.copy(ref_counter_chains)

            # Resetting the flag (run) to False. If there are still chains to be added, it is turned to True, and the while loop
            # continues
            run = False
            for ch, n in ref_counter.items():
                if n != 0:
                    run = True
                    break

        if clash_count >= threshold_clashes:
            if options.version == "short":
                sys.stderr.write("Try the intensive version (-ver or --version intensive) or select another clash cut-off (-cc or --clash_distance_cutoff).\n")
                exit()
        else:

            # Save the final model (with all the chains). In case the model has more than 83 chains, another model is created
            # to avoid problems with chain ids.
            print("\nSAVING MODEL %d...\n" % n_model)

            # Adding only current_model if it is not yet in the list_models
            list_models.append(current_model)
            set_models = set(list_models)

            # Creating a subdirectory (outputs) of the models directory, for each one of the models.
            outputs = (output_models + str(n_model) + "/")
            if not os.path.exists(outputs):
                os.makedirs(outputs)

            mod = 0
            for model in set_models:
                model_filename = n_model
                pdb_model_filename = outputs + str(model_filename) + "_" + str(mod) + ".aligned.pdb"
                io = Bio.PDB.PDBIO()
                io.set_structure(model)
                io.save(pdb_model_filename)
                mod += 1

            print("The model '%s' has been saved here:\n %s\n\n" % (pdb_model_filename, outputs))

            if options.energy_plot:
                # Preparing for the energy analysis (Modeller) -> creating outputs/temp/_mod.pdb
                temp_dir = create_tempPDB(outputs)

                # DOPE profile plot for each model
                blockPrint()
                DOPE_energy_plot=DOPE_profiles_maker(temp_dir, outputs)
                enablePrint()
                print(DOPE_energy_plot)

            if options.refine:
                print("Starting to refine the models. This step may take a few minutes...\n")

                refined_models = outputs + "optimization_results/refined_models/"
                if not os.path.exists(refined_models):
                    os.makedirs(refined_models)

                stats = outputs + "optimization_results/stats/"
                if not os.path.exists(stats):
                    os.makedirs(stats)

                dope_profile = outputs + "optimization_results/dope_profile/"
                if not os.path.exists(dope_profile):
                    os.makedirs(dope_profile)

                log_files = outputs + "optimization_results/log_files/"
                if not os.path.exists(log_files):
                    os.makedirs(log_files)

                if options.verbose:
                    sys.stderr.write("The following directories will be created:\n  %s\n  %s\n  %s\n  %s\n\n" % (refined_models, stats, dope_profile, log_files))

                refine_model(outputs, log_files, stats, refined_models, dope_profile)

                models_dir = create_tempPDB(outputs)

                refined_comparison = refined_vs_notrefined(models_dir, dope_profile)
                print(refined_comparison)

                if options.verbose:
                    sys.stderr.write("You can find detailed information about the refining process in the log file.\n")

        if options.version != "short":
            continue
        else:
            break

print("\n\nTHE CONSTRUCTION PROCESS HAS ENDED\n")