Question

I am trying to convert some matlab code from the Maia package into something that will work with Octave. I am currently getting stuck because one of the files has several calls to containers.Map which is apparently something that has not yet been implemented in octave. Does anyone have any ideas for easily achieving similar functionality without doing a whole lot of extra work in octave? Thanks all for your time.

function [adj_direct contig_direct overlap names longest_path_direct...
          weigth_direct deltafiles deltafiles_ref ReferenceAlignment ...
          contig_ref overlap_ref name_hash_ref] = ...
          assembly_driver(assemblies,ref_genome,target_chromosome, ... 
                          deltafiles_ref,contig_ref, overlap_ref, ...
                          name_hash_ref, varargin)

% ASSEMBLY_DRIVER Combines contig sets into one assembled chromosome
%
% INPUT 
%   assemblies
%   ref_chromosome
%   Startnode_name
%   Endnode_name
%        OPTIONAL              DEFAULT
%       'z_weigths'            [.25 .25 .25 .25]
%       'clipping_thrs'        10
%       'ref_distance'         -10
%       'ref_quality'          1E-5
%       'max_chromosome_dist'  100
%       'quit_treshold'        15
%       'tabu_time'            3
%       'minimum_improvement'  -inf
%       'ref_node_assemblies'  all assemblies (slow)
%       'endextend'            true
%
%

    % SET DEFAULTS
    % General parameters
    z_weights           = [.25 .25 .25 .25];
    clipping_thrs       = 10;
    mapfilter           = '-rq';
    alignlen            = 75;
    ident               = 85;

    % Reference nod parameters
    ref_distance        = -10;
    ref_quality         = 1E-5;
    max_chromosome_dist = 100;
    % TABU parameters
    quit_treshold       = 15;
    tabu_time           = 3;
    minimum_improvement = -inf;
    ref_node_assemblies = assemblies;
    % Extending the assembly outwards from the start and en node
    endextend           = true;
    AllowReverse        = true;
    % If no start and end node are given, they will be determined from tiling
    Startnode_name      = '';
    Endnode_name        = '';
    containment_edge    = true;
    ref_first           = true;

    % If contigs have already been aligned to the reference, give the
    % deltafile 
    ReferenceAlignment = 'NotYetDoneByMaia';

    % Get VARARGIN user input
    if length(varargin) > 0
        while 1
            switch varargin{1}
                case 'Startnode_name'
                    Startnode_name = varargin{2};
                case 'Endnode_name'
                    Endnode_name = varargin{2};
                case 'z_weigths'
                    z_weights = varargin{2};
                case 'clipping_thrs' 
                    clipping_thrs = varargin{2};
                case 'ref_distance' 
                    ref_distance = varargin{2};
                case 'ref_quality' 
                    ref_quality = varargin{2};
                case 'max_chromosome_dist' 
                    max_chromosome_dist = varargin{2};
                case 'quit_treshold' 
                    quit_treshold = varargin{2};
                case 'tabu_time' 
                    tabu_time = varargin{2};
                case 'minimum_improvement' 
                    minimum_improvement = varargin{2};
                case 'ref_node_assemblies'
                    ref_node_assemblies = assemblies(varargin{2},:);
                case 'extend_ends'
                    endextend = assemblies(varargin{2},:);
                case 'AllowReverse'
                    AllowReverse = varargin{2};
                case 'ReferenceAlignment'
                    ReferenceAlignment = varargin{2};
                case 'containment_edge'
                    containment_edge = varargin{2};
                case 'ref_first'
                    ref_first = varargin{2};
                case 'mapfilter'
                    mapfilter = varargin{2};
                case 'alignlen'
                    alignlen = varargin{2};
                case 'ident'
                    ident = varargin{2};
                otherwise
                    error(['Input ' varargin{2} ' is not known']);
            end
            if length(varargin) > 2
                varargin = varargin(3:end);
            else
                break;
            end
        end
    end

    % Read input assemblies
    assembly_names   = assemblies(:,1);
    assembly_locs    = assemblies(:,2);
    assembly_quality = containers.Map(assemblies(:,1),assemblies(:,3));
    assembly_quality('reference') = ref_quality;

    % Read input assemblies for creation of reference nodes
    ref_node_assembly_names   = ref_node_assemblies(:,1);
    ref_node_assembly_locs    = ref_node_assemblies(:,2);
    ref_node_assembly_quality = containers.Map(ref_node_assemblies(:,1),ref_node_assemblies(:,3));
    ref_node_assembly_quality('reference') = ref_quality;


    % If there is only one assembly there is nothing to align
    if size(assemblies,1) >= 2

        % Align assemblies against each other
        assembly_pairs = {};
        coordsfiles = [];
        deltafiles = [];
        for i = 1:length(assembly_locs)-1
            for j = i+1:length(assembly_locs)
                [coordsfile,deltafile] = align_assemblies({assembly_locs{i},assembly_locs{j}},{assembly_names{i}, assembly_names{j}}, ...
                                                           mapfilter, alignlen, ident);
                coordsfiles = [coordsfiles; coordsfile];
                %deltafiles = [deltafiles deltafile];
                deltafiles = [deltafiles; {deltafile}];
                assembly_pairs = [assembly_pairs;[assembly_names(i) assembly_names(j)]];
            end
        end


     %   fprintf('Loading alignment files.\n');
     %   load alignments_done;

        % Put the nucmer alignments in an adjency matrix
        %[adj, names, name_hash, contig, overlap] = get_adj_matrix(coordsfiles, assembly_pairs, assembly_quality, z_weights, 'clipping_thrs', clipping_thrs, 'dove_tail', 'double','edge_weight','z-scores', 'containment_edge', true);
        [adj, names, name_hash, contig, overlap] = get_adj_matrix(deltafiles, assembly_pairs, assembly_quality, z_weights, 'clipping_thrs', clipping_thrs, 'dove_tail', 'double','edge_weight','z-scores', 'containment_edge', containment_edge);



        % Merge deltafiles
        deltafilesnew = deltafiles{1};
        if size(deltafiles,1) > 1
            for di = 2:size(deltafiles,1)
                deltafilesnew = [deltafilesnew deltafiles{di}];
            end
        end
        deltafiles = deltafilesnew;

    else
        assembly_pairs = {};
        coordsfiles = [];
        deltafiles = [];        
        adj = [];
        names = {};
        name_hash = containers.Map;
        contig  = struct('name',{},'size',[],'chromosome',[],'number',[], 'assembly', [], 'assembly_quality', []);
        overlap = struct('Q',{},'R',[],'S1',[],'E1', [], 'S2', [], 'E2', [], 'LEN1', [], 'LEN2', [], 'IDY', [], 'COVR', [], 'COVQ', [],'LENR',[], 'LENQ',[]);
    end


    % Ad the pseudo nodes to the graph. If the contigs have already been
    % aligned to the reference genome, just select the alignments that
    % correspond to the target chromosome
    if isequal(ReferenceAlignment,'NotYetDoneByMaia')
        % Align all contigs in 'contig_sets_fasta' to the reference chromosome 
        [contig_ref, overlap_ref, name_hash_ref, deltafiles_ref] = align_contigs_sets(... 
            ref_genome, ref_node_assembly_locs, ref_node_assembly_names, ... 
            ref_node_assembly_quality, clipping_thrs, z_weights, ... 
            ref_distance,max_chromosome_dist);

        ReferenceAlignment = 'out2.delta';
    end
    % Select only the entries in the deltafile for the current target chromosome
    [contig_target_ref, overlap_target_ref, name_hash_target_ref, delta_target_ref] = ...
              GetVariablesForTargetChromosome(...
              contig_ref, overlap_ref, deltafiles_ref);


    % Ref clipping should be high in case of tiling
    %if isequal(max_chromosome_dist,'tiling')
    %    clipping_thrs = 10000
    %end

    % Add reference nodes to the adjency matrix
    [adj, names, name_hash, contig, overlap, delta_target_ref, Startnode_name, Endnode_name] = get_reference_nodes( ...
                     adj, names, name_hash, contig, overlap, target_chromosome, ...
                     contig_target_ref, overlap_target_ref, name_hash_target_ref, delta_target_ref, ...
                     max_chromosome_dist, ref_distance, clipping_thrs, ref_first,...
                     Startnode_name, Endnode_name, AllowReverse);


    % Give reference edges some small extra value to distict between
    % assemblies to which a reference node leads
    % adj = rank_reference_edges(adj,contig,assembly_quality);

    % Specify a start and an end node for the assembly
    Startnode = name_hash(Startnode_name);
    Endnode = name_hash(Endnode_name);


    % Find the best scoring path
    fprintf('Directing the final graph\n');
    % Calculate path on undirected graph to get an idea on how to direct the graph
    [longest_path weigth] = longest_path_tabu(adj, Startnode, Endnode, quit_treshold, tabu_time, minimum_improvement);
    % Make the graph directed (greedy)
    [adj_direct contig_direct] = direct_graph(adj,overlap, contig, names, name_hash,clipping_thrs, Startnode, longest_path, true, ref_first);
    % Calcultate final layout-path
    fprintf('Find highest scoring path\n');
    [longest_path_direct weigth_direct] = longest_path_tabu(adj_direct, Startnode, Endnode, quit_treshold, tabu_time, minimum_improvement);


    function [contig_target_ref, overlap_target_ref, name_hash_target_ref, delta_target_ref] = ...
              GetVariablesForTargetChromosome(...
              contig_ref, overlap_ref, deltafiles_ref)

        % Select only the entries in the deltafile for the current target chromosome
        delta_target_ref = deltafiles_ref;
        for di = size(delta_target_ref,2):-1:1
            if ~isequal(delta_target_ref(di).R,target_chromosome)
                delta_target_ref(di) = [];
            end
        end
        overlap_target_ref = overlap_ref;
        for oi = size(overlap_target_ref,2):-1:1
            if ~isequal(overlap_target_ref(oi).R,target_chromosome)
                overlap_target_ref(oi) = [];
            end
        end    
        contig_target_ref = contig_ref;
        for ci = size(contig_target_ref,1):-1:1
            if isequal(contig_target_ref(ci).assembly, 'reference') && ~isequal(contig_target_ref(ci).name,target_chromosome)
                contig_target_ref(ci) = [];
            end
        end    
        name_hash_target_ref = make_hash({contig_target_ref.name}');
    end


end

Answer 1

There is no exact equivalent of containers.Map in Octave that I know of…

One option is to use the java package to create java.util.Hashtable. Using this example:

pkg load java
d = javaObject("java.util.Hashtable");
d.put('a',1)
d.put('b',2)
d.put('c',3)
d.get('b')

---

If you are willing to do a bit of rewriting, you can use the builtin struct as a rudimentary hash table with strings (valid variable names) as keys, and pretty much anything stored in values.

For example, given the following:

keys = {'Mon','Tue','Wed'}
values = {10, 20, 30}

you could replace this:

map = containers.Map(keys,values);
map('Mon')

by:

s = struct();
for i=1:numel(keys)
    s.(keys{i}) = values{i};
end
s.('Mon')

You might need to use genvarname to produce valid keys, or maybe a proper hashing function that produces valid key strings.

Also look into struct-related functions: getfield, setfield, isfield, fieldnames, rmfield, etc..