molecular_network_library.py

#!/usr/bin/python

import ming_fileio_library
import ming_proteosafe_library

"""
Molecular Network Utilties For Use at GNPS

These classes provide utilies for loading and doing manipulation of the files

Refined Version

"""


class NetworkPair:
    def __init__(self, node1, node2, cosine, deltamz):
        self.node1 = node1
        self.node2 = node2
        self.cosine = cosine
        self.deltamz = deltamz

class OriginalSpectrum:
    def __init__(self, mz, charge, clusterindex, filename, scan):
        self.mz = mz
        self.charge = charge
        self.clusterindex = clusterindex
        self.filename = filename
        self.scan = scan

class ClusterLibraryIdentification:
    def __init__(self, spectrumID, name, smiles, inchi, score, scan):
        self.spectrumID = spectrumID
        self.name = name
        self.smiles = smiles
        self.inchi = inchi
        self.score = score
        self.scan = scan

class ClusterPeptideIdentification:
    def __init__(self, peptide, score, engine):
        self.peptide = peptide
        self.score = score
        self.engine = engine

class ClusterNode:
    def __init__(self, mz, charge, index, number_of_spectra, component_index):
        self.mz = mz
        self.charge = charge
        self.index = index
        self.component = component_index
        self.number_of_spectra = number_of_spectra
        self.constituent_spectra = []
        self.connected_pairs = []
        self.library_identifications = []
        self.all_files_string = ""

    def add_cluster_spectrum(self, filename, scan):
        self.constituent_spectra.append(filename + ":" + scan)

    def add_connected_pair(self, pair_object):
        self.connected_pairs.append(pair_object)

    def get_node_files(self):
        list_files = []
        for token in self.all_files_string.split("###"):
            if len(token.split(":")[0]) > 0:
                list_files.append(token.split(":")[0])
        return list_files

    #Returns whether the node has been identified
    def is_identified(self):
        if len(self.library_identifications) == 0:
            return False
        else:
            return True

    def does_contain_constituent_spectra(self, filename, scan):
        spectrum_key = filename + ":" + scan
        if self.constituent_spectra.count(spectrum_key) > 0:
            return True
        return False


class NetworkConnectedComponent:
    def __init__(self, component_index):
        self.component_index = component_index
        self.nodes = []

    def add_node_to_component(self, clusternode_object):
        self.nodes.append(clusternode_object)

class MolecularNetwork:
    def __init__(self):
        self.nodes = []
        self.index_to_node_map = {} #Optimization for fast access by node index
        self.index_to_neighbors = {} #Key is scan, and points to list of node objects that are neighbors
        self.pairs = []
        self.identifications = []
        self.filemapping = {} 	#Mapping from mangled name to real filename

    def load_network(self, clustersummaryfilename, pairs_filename):
        self.load_clustersummary(clustersummaryfilename)
        self.load_pairsinfo(pairs_filename)

        #Making things consistent after loading individual files

    ###
     # Must be loaded first
    ###
    def load_clustersummary(self, clustersummaryfilename):
        row_count, table_data = ming_fileio_library.parse_table_with_headers(clustersummaryfilename)

        for i in range(row_count):
            cluster_index = table_data["cluster index"][i]
            mz = table_data["precursor mass"][i]
            charge = table_data["precursor charge"][i]
            parentmass = table_data["parent mass"][i]
            number_of_spectra = table_data["number of spectra"][i]
            all_files = table_data["AllFiles"][i]

            componentindex = -1
            if "componentindex" in table_data:
                componentindex = table_data["componentindex"][i]

            cluster_node = ClusterNode(mz, charge, cluster_index, number_of_spectra, componentindex)
            cluster_node.all_files_string = all_files

            self.nodes.append(cluster_node)
            self.index_to_node_map[cluster_index] = cluster_node

            #Making all the nodes not shit in terms of clustering info
            constituent_spectra = cluster_node.all_files_string.split("###")
            cluster_node.constituent_spectra = constituent_spectra


        #Make stuff consistent with components


    def load_parameters_file(self, paramsfilename):
        #Loading the file mapping
        parameters = ming_proteosafe_library.parse_xml_file(open(paramsfilename, "r"))
        mangled_mapping = ming_proteosafe_library.get_mangled_file_mapping(parameters)
        self.mangled_mapping = mangled_mapping


    ###
     # Loading the pairs info
     # Requires clustersummary to be loaded first
    ###
    def load_pairsinfo(self, pairs_filename):
        row_count, table_data = ming_fileio_library.parse_table_with_headers(pairs_filename)

        if "CLUSTERID1" in table_data:
            for i in range(row_count):
                node1 = table_data["CLUSTERID1"][i]
                node2 = table_data["CLUSTERID2"][i]
                cosine = table_data["Cosine"][i]
                deltamz = table_data["DeltaMZ"][i]
                pair = NetworkPair(node1, node2, cosine, deltamz)
                self.pairs.append(pair)
        else:
            row_count, table_data = ming_fileio_library.parse_table_without_headers(pairs_filename)
            for i in range(row_count):
                node1 = table_data[0][i]
                node2 = table_data[1][i]
                cosine = table_data[4][i]
                deltamz = table_data[2][i]
                pair = NetworkPair(node1, node2, cosine, deltamz)
                self.pairs.append(pair)



        #Make stuff consistent, specifically adding adjacency list
        for pair in self.pairs:
            node1 = pair.node1
            node2 = pair.node2

            if not(node1 in self.index_to_neighbors):
                self.index_to_neighbors[node1] = []

            if not(node2 in self.index_to_neighbors):
                self.index_to_neighbors[node2] = []

            self.index_to_neighbors[node1].append(node2)
            self.index_to_neighbors[node2].append(node1)

    #Loading the identifications of the clusters by library search
    def load_gnps_librarysearch(self, identification_filename):
        row_count, table_data = ming_fileio_library.parse_table_with_headers(identification_filename)
        for i in range(row_count):
            compound_name = table_data["Compound_Name"][i]
            smiles = table_data["Smiles"][i]
            inchi = table_data["INCHI"][i]
            SpectrumID = table_data["SpectrumID"][i]
            score = table_data["MQScore"][i]
            scan = table_data["#Scan#"][i]
            identification = ClusterLibraryIdentification(SpectrumID, compound_name, smiles, inchi, score, scan)
            self.identifications.append(identification)

            #Finding the cluster
            if scan in self.index_to_node_map:
                self.index_to_node_map[scan].library_identifications.append(identification)

    #Getter Functions
    def get_node_count(self):
        return len(self.nodes)

    #Get total number of raw spectra that was in consideration for networking
    def get_cluster_total_spectrum_counts(self):
        total_count = 0
        for node in self.nodes:
            total_count += int(node.number_of_spectra)
        return total_count

    #Returns a list of all the filenames in the network
    def get_files_list_in_clusters(self):
        file_list = []
        for node in self.nodes:
            file_list += node.get_node_files()
        return list(set(file_list))

    #Getting the Cluster Information by Index
    def get_cluster_index(self, scan):
        if scan in self.index_to_node_map:
            return self.index_to_node_map[scan]
        return None

    def get_node_neighbors(self, scan):
        if scan in self.index_to_neighbors:
            return self.index_to_neighbors[scan]
        else:
            return []

    def get_unidentified_node_neighbors(self, scan):
        if scan in self.index_to_neighbors:
            return_nodes = []
            for node_id in self.index_to_neighbors[scan]:
                if not self.get_cluster_index(node_id).is_identified():
                    return_nodes.append(node_id)
        else:
            return []

        return return_nodes

    def get_identified_node_neighbors(self, scan):
        if scan in self.index_to_neighbors:
            return_nodes = []
            for node_id in self.index_to_neighbors[scan]:
                if self.get_cluster_index(node_id).is_identified():
                    return_nodes.append(node_id)
        else:
            return []

        return return_nodes

    def get_node_from_raw_data(self, raw_filename, scan_number):
        for node in self.nodes:
            if node.does_contain_constituent_spectra(raw_filename, scan_number):
                return node

        return None