import java.io.Writer;
import java.util.ArrayList;
import java.util.Date;
+import java.util.HashMap;
import java.util.HashSet;
import java.util.List;
import java.util.Map;
+import java.util.Map.Entry;
import java.util.Set;
import java.util.SortedMap;
import java.util.SortedSet;
import org.forester.evoinference.distance.NeighborJoining;
import org.forester.evoinference.matrix.character.CharacterStateMatrix.Format;
-import org.forester.evoinference.matrix.distance.DistanceMatrix;
+import org.forester.evoinference.matrix.distance.BasicSymmetricalDistanceMatrix;
import org.forester.go.GoId;
import org.forester.go.GoNameSpace;
import org.forester.go.GoTerm;
import org.forester.go.PfamToGoParser;
import org.forester.io.parsers.HmmscanPerDomainTableParser;
import org.forester.io.parsers.HmmscanPerDomainTableParser.INDIVIDUAL_SCORE_CUTOFF;
+import org.forester.io.parsers.util.ParserUtils;
import org.forester.io.writers.PhylogenyWriter;
import org.forester.phylogeny.Phylogeny;
import org.forester.phylogeny.PhylogenyMethods;
import org.forester.phylogeny.PhylogenyNode;
import org.forester.phylogeny.factories.ParserBasedPhylogenyFactory;
import org.forester.phylogeny.iterators.PhylogenyNodeIterator;
+import org.forester.protein.BinaryDomainCombination;
+import org.forester.protein.Domain;
+import org.forester.protein.DomainId;
+import org.forester.protein.Protein;
+import org.forester.species.BasicSpecies;
+import org.forester.species.Species;
import org.forester.surfacing.BasicDomainSimilarityCalculator;
import org.forester.surfacing.BasicGenomeWideCombinableDomains;
-import org.forester.surfacing.BasicSpecies;
-import org.forester.surfacing.BinaryDomainCombination;
import org.forester.surfacing.CombinationsBasedPairwiseDomainSimilarityCalculator;
import org.forester.surfacing.DomainCountsBasedPairwiseSimilarityCalculator;
import org.forester.surfacing.DomainCountsDifferenceUtil;
-import org.forester.surfacing.DomainId;
import org.forester.surfacing.DomainLengthsTable;
import org.forester.surfacing.DomainParsimonyCalculator;
import org.forester.surfacing.DomainSimilarity;
import org.forester.surfacing.PairwiseGenomeComparator;
import org.forester.surfacing.PrintableDomainSimilarity;
import org.forester.surfacing.PrintableDomainSimilarity.PRINT_OPTION;
-import org.forester.surfacing.Protein;
import org.forester.surfacing.ProteinCountsBasedPairwiseDomainSimilarityCalculator;
-import org.forester.surfacing.Species;
import org.forester.surfacing.SurfacingUtil;
+import org.forester.util.BasicDescriptiveStatistics;
import org.forester.util.BasicTable;
import org.forester.util.BasicTableParser;
import org.forester.util.CommandLineArguments;
public class surfacing {
- public final static String DOMAIN_COMBINITONS_OUTPUT_OPTION_FOR_GRAPH_ANALYSIS = "graph_analysis_out";
- public final static String DOMAIN_COMBINITONS_OUTPUTFILE_SUFFIX_FOR_GRAPH_ANALYSIS = "_dc.dot";
- public final static String PARSIMONY_OUTPUT_FITCH_PRESENT_BC_OUTPUTFILE_SUFFIX_FOR_GRAPH_ANALYSIS = "_fitch_present_dc.dot";
- public final static String DOMAIN_COMBINITON_COUNTS_OUTPUTFILE_SUFFIX = ".dcc";
+ private static final int MINIMAL_NUMBER_OF_SIMILARITIES_FOR_SPLITTING = 1000;
+ public final static String DOMAIN_COMBINITONS_OUTPUT_OPTION_FOR_GRAPH_ANALYSIS = "graph_analysis_out";
+ public final static String DOMAIN_COMBINITONS_OUTPUTFILE_SUFFIX_FOR_GRAPH_ANALYSIS = "_dc.dot";
+ public final static String PARSIMONY_OUTPUT_FITCH_PRESENT_BC_OUTPUTFILE_SUFFIX_FOR_GRAPH_ANALYSIS = "_fitch_present_dc.dot";
+ public final static String DOMAIN_COMBINITON_COUNTS_OUTPUTFILE_SUFFIX = ".dcc";
// gain/loss:
- public final static String PARSIMONY_OUTPUT_GL_SUFFIX_DOLLO_DOMAINS = "_dollo_gl_d";
- public final static String PARSIMONY_OUTPUT_GL_SUFFIX_DOLLO_BINARY_COMBINATIONS = "_dollo_gl_dc";
- public final static String PARSIMONY_OUTPUT_GL_SUFFIX_FITCH_DOMAINS = "_fitch_gl_d";
- public final static String PARSIMONY_OUTPUT_GL_SUFFIX_FITCH_BINARY_COMBINATIONS = "_fitch_gl_dc";
+ public final static String PARSIMONY_OUTPUT_GL_SUFFIX_DOLLO_DOMAINS = "_dollo_gl_d";
+ public final static String PARSIMONY_OUTPUT_GL_SUFFIX_DOLLO_BINARY_COMBINATIONS = "_dollo_gl_dc";
+ public final static String PARSIMONY_OUTPUT_GL_SUFFIX_FITCH_DOMAINS = "_fitch_gl_d";
+ public final static String PARSIMONY_OUTPUT_GL_SUFFIX_FITCH_BINARY_COMBINATIONS = "_fitch_gl_dc";
// gain/loss counts:
- public final static String PARSIMONY_OUTPUT_GL_COUNTS_SUFFIX_DOLLO_DOMAINS = "_dollo_glc_d";
- public final static String PARSIMONY_OUTPUT_GL_COUNTS_SUFFIX_DOLLO_BINARY_COMBINATIONS = "_dollo_glc_dc";
- public final static String PARSIMONY_OUTPUT_GL_COUNTS_SUFFIX_FITCH_DOMAINS = "_fitch_glc_d";
- public final static String PARSIMONY_OUTPUT_GL_COUNTS_SUFFIX_FITCH_BINARY_COMBINATIONS = "_fitch_glc_dc";
+ public final static String PARSIMONY_OUTPUT_GL_COUNTS_SUFFIX_DOLLO_DOMAINS = "_dollo_glc_d";
+ public final static String PARSIMONY_OUTPUT_GL_COUNTS_SUFFIX_DOLLO_BINARY_COMBINATIONS = "_dollo_glc_dc";
+ public final static String PARSIMONY_OUTPUT_GL_COUNTS_SUFFIX_FITCH_DOMAINS = "_fitch_glc_d";
+ public final static String PARSIMONY_OUTPUT_GL_COUNTS_SUFFIX_FITCH_BINARY_COMBINATIONS = "_fitch_glc_dc";
// tables:
- public final static String PARSIMONY_OUTPUT_FITCH_GAINS_BC = "_fitch_gains_dc";
- public final static String PARSIMONY_OUTPUT_FITCH_GAINS_HTML_BC = "_fitch_gains_dc.html";
- public final static String PARSIMONY_OUTPUT_FITCH_LOSSES_BC = "_fitch_losses_dc";
- public final static String PARSIMONY_OUTPUT_FITCH_LOSSES_HTML_BC = "_fitch_losses_dc.html";
- public final static String PARSIMONY_OUTPUT_FITCH_PRESENT_BC = "_fitch_present_dc";
- public final static String PARSIMONY_OUTPUT_FITCH_PRESENT_HTML_BC = "_fitch_present_dc.html";
- public final static String PARSIMONY_OUTPUT_DOLLO_GAINS_D = "_dollo_gains_d";
- public final static String PARSIMONY_OUTPUT_DOLLO_GAINS_HTML_D = "_dollo_gains_d.html";
- public final static String PARSIMONY_OUTPUT_DOLLO_LOSSES_D = "_dollo_losses_d";
- public final static String PARSIMONY_OUTPUT_DOLLO_LOSSES_HTML_D = "_dollo_losses_d.html";
- public final static String PARSIMONY_OUTPUT_DOLLO_PRESENT_D = "_dollo_present_d";
- public final static String PARSIMONY_OUTPUT_DOLLO_PRESENT_HTML_D = "_dollo_present_d.html";
- public final static String DOMAINS_PRESENT_NEXUS = "_dom.nex";
- public final static String BDC_PRESENT_NEXUS = "_dc.nex";
+ public final static String PARSIMONY_OUTPUT_FITCH_GAINS_BC = "_fitch_gains_dc";
+ public final static String PARSIMONY_OUTPUT_FITCH_GAINS_HTML_BC = "_fitch_gains_dc.html";
+ public final static String PARSIMONY_OUTPUT_FITCH_LOSSES_BC = "_fitch_losses_dc";
+ public final static String PARSIMONY_OUTPUT_FITCH_LOSSES_HTML_BC = "_fitch_losses_dc.html";
+ public final static String PARSIMONY_OUTPUT_FITCH_PRESENT_BC = "_fitch_present_dc";
+ public final static String PARSIMONY_OUTPUT_FITCH_PRESENT_HTML_BC = "_fitch_present_dc.html";
+ public final static String PARSIMONY_OUTPUT_DOLLO_GAINS_D = "_dollo_gains_d";
+ public final static String PARSIMONY_OUTPUT_DOLLO_GAINS_HTML_D = "_dollo_gains_d.html";
+ public final static String PARSIMONY_OUTPUT_DOLLO_LOSSES_D = "_dollo_losses_d";
+ public final static String PARSIMONY_OUTPUT_DOLLO_LOSSES_HTML_D = "_dollo_losses_d.html";
+ public final static String PARSIMONY_OUTPUT_DOLLO_PRESENT_D = "_dollo_present_d";
+ public final static String PARSIMONY_OUTPUT_DOLLO_PRESENT_HTML_D = "_dollo_present_d.html";
+ public final static String DOMAINS_PRESENT_NEXUS = "_dom.nex";
+ public final static String BDC_PRESENT_NEXUS = "_dc.nex";
// ---
- public final static String PRG_NAME = "surfacing";
- public static final String DOMAINS_PARSIMONY_TREE_OUTPUT_SUFFIX_DOLLO = "_d_dollo"
- + ForesterConstants.PHYLO_XML_SUFFIX;
- public static final String DOMAINS_PARSIMONY_TREE_OUTPUT_SUFFIX_FITCH = "_d_fitch"
- + ForesterConstants.PHYLO_XML_SUFFIX;
- public static final String BINARY_DOMAIN_COMBINATIONS_PARSIMONY_TREE_OUTPUT_SUFFIX_DOLLO = "_dc_dollo"
- + ForesterConstants.PHYLO_XML_SUFFIX;
- public static final String BINARY_DOMAIN_COMBINATIONS_PARSIMONY_TREE_OUTPUT_SUFFIX_FITCH = "_dc_fitch"
- + ForesterConstants.PHYLO_XML_SUFFIX;
- public static final String NEXUS_EXTERNAL_DOMAINS = "_dom.nex";
- public static final String NEXUS_EXTERNAL_DOMAIN_COMBINATIONS = "_dc.nex";
- public static final String NEXUS_SECONDARY_FEATURES = "_secondary_features.nex";
- public static final String PARSIMONY_OUTPUT_GL_SUFFIX_DOLLO_SECONDARY_FEATURES = "_dollo_gl_secondary_features";
- public static final String PARSIMONY_OUTPUT_GL_COUNTS_SUFFIX_DOLLO_SECONDARY_FEATURES = "_dollo_glc_secondary_features";
- public static final String PARSIMONY_OUTPUT_DOLLO_GAINS_SECONDARY_FEATURES = "_dollo_gains_secondary_features";
- public static final String PARSIMONY_OUTPUT_DOLLO_LOSSES_SECONDARY_FEATURES = "_dollo_losses_secondary_features";
- public static final String PARSIMONY_OUTPUT_DOLLO_PRESENT_SECONDARY_FEATURES = "_dollo_present_secondary_features";
- public static final String SECONDARY_FEATURES_PARSIMONY_TREE_OUTPUT_SUFFIX_DOLLO = "_secondary_features_dollo"
- + ForesterConstants.PHYLO_XML_SUFFIX;
- public static final String PARSIMONY_OUTPUT_DOLLO_ALL_GOID_D_ALL_NAMESPACES = "_dollo_goid_d";
- public static final String PARSIMONY_OUTPUT_FITCH_ALL_GOID_BC_ALL_NAMESPACES = "_fitch_goid_dc";
- final static private String HELP_OPTION_1 = "help";
- final static private String HELP_OPTION_2 = "h";
- final static private String OUTPUT_DIR_OPTION = "out_dir";
- final static private String SCORING_OPTION = "scoring";
- private static final DomainSimilarityScoring SCORING_DEFAULT = DomainSimilarity.DomainSimilarityScoring.COMBINATIONS;
- final static private String SCORING_DOMAIN_COUNT_BASED = "domains";
- final static private String SCORING_PROTEIN_COUNT_BASED = "proteins";
- final static private String SCORING_COMBINATION_BASED = "combinations";
- final static private String DETAILEDNESS_OPTION = "detail";
- private final static Detailedness DETAILEDNESS_DEFAULT = DomainSimilarityCalculator.Detailedness.PUNCTILIOUS;
- final static private String SPECIES_MATRIX_OPTION = "smatrix";
- final static private String DETAILEDNESS_BASIC = "basic";
- final static private String DETAILEDNESS_LIST_IDS = "list_ids";
- final static private String DETAILEDNESS_PUNCTILIOUS = "punctilious";
- final static private String DOMAIN_SIMILARITY_SORT_OPTION = "sort";
- private static final DomainSimilaritySortField DOMAIN_SORT_FILD_DEFAULT = DomainSimilarity.DomainSimilaritySortField.DOMAIN_ID;
- final static private String DOMAIN_SIMILARITY_SORT_MIN = "min";
- final static private String DOMAIN_SIMILARITY_SORT_MAX = "max";
- final static private String DOMAIN_SIMILARITY_SORT_SD = "sd";
- final static private String DOMAIN_SIMILARITY_SORT_MEAN = "mean";
- final static private String DOMAIN_SIMILARITY_SORT_DIFF = "diff";
- final static private String DOMAIN_SIMILARITY_SORT_COUNTS_DIFF = "count_diff";
- final static private String DOMAIN_SIMILARITY_SORT_ABS_COUNTS_DIFF = "abs_count_diff";
- final static private String DOMAIN_SIMILARITY_SORT_SPECIES_COUNT = "species";
- final static private String DOMAIN_SIMILARITY_SORT_ALPHA = "alpha";
- final static private String DOMAIN_SIMILARITY_SORT_BY_SPECIES_COUNT_FIRST_OPTION = "species_first";
- final static private String DOMAIN_COUNT_SORT_OPTION = "dc_sort";
- private static final GenomeWideCombinableDomainsSortOrder DOMAINS_SORT_ORDER_DEFAULT = GenomeWideCombinableDomains.GenomeWideCombinableDomainsSortOrder.ALPHABETICAL_KEY_ID;
- final static private String DOMAIN_COUNT_SORT_ALPHA = "alpha";
- final static private String DOMAIN_COUNT_SORT_KEY_DOMAIN_COUNT = "dom";
- final static private String DOMAIN_COUNT_SORT_KEY_DOMAIN_PROTEINS_COUNT = "prot";
- final static private String DOMAIN_COUNT_SORT_COMBINATIONS_COUNT = "comb";
- final static private String CUTOFF_SCORE_FILE_OPTION = "cos";
- final static private String NOT_IGNORE_DUFS_OPTION = "dufs";
- final static private String MAX_E_VALUE_OPTION = "e";
- final static private String MAX_ALLOWED_OVERLAP_OPTION = "mo";
- final static private String NO_ENGULFING_OVERLAP_OPTION = "no_eo";
- final static private String IGNORE_COMBINATION_WITH_SAME_OPTION = "ignore_self_comb";
- final static private String PAIRWISE_DOMAIN_COMPARISONS_PREFIX = "pwc_";
- final static private String PAIRWISE_DOMAIN_COMPARISONS_OPTION = "pwc";
- final static private String OUTPUT_FILE_OPTION = "o";
- final static private String PFAM_TO_GO_FILE_USE_OPTION = "p2g";
- final static private String GO_OBO_FILE_USE_OPTION = "obo";
- final static private String GO_NAMESPACE_LIMIT_OPTION = "go_namespace";
- final static private String GO_NAMESPACE_LIMIT_OPTION_MOLECULAR_FUNCTION = "molecular_function";
- final static private String GO_NAMESPACE_LIMIT_OPTION_BIOLOGICAL_PROCESS = "biological_process";
- final static private String GO_NAMESPACE_LIMIT_OPTION_CELLULAR_COMPONENT = "cellular_component";
- final static private String SECONDARY_FEATURES_PARSIMONY_MAP_FILE = "secondary";
- final static private String DOMAIN_SIMILARITY_PRINT_OPTION_SIMPLE_TAB_DELIMITED = "simple_tab";
- final static private String DOMAIN_SIMILARITY_PRINT_OPTION_SIMPLE_HTML = "simple_html";
- final static private String DOMAIN_SIMILARITY_PRINT_OPTION_DETAILED_HTML = "detailed_html";
- final static private String DOMAIN_SIMILARITY_PRINT_OPTION = "ds_output";
- private static final PRINT_OPTION DOMAIN_SIMILARITY_PRINT_OPTION_DEFAULT = PrintableDomainSimilarity.PRINT_OPTION.HTML;
- final static private String IGNORE_DOMAINS_WITHOUT_COMBINATIONS_IN_ALL_SPECIES_OPTION = "ignore_singlet_domains";
- final static private String IGNORE_VIRAL_IDS = "ignore_viral_ids";
- final static private boolean IGNORE_DOMAINS_WITHOUT_COMBINATIONS_IN_ALL_SPECIES_DEFAULT = false;
- final static private String IGNORE_DOMAINS_SPECIFIC_TO_ONE_SPECIES_OPTION = "ignore_species_specific_domains";
- final static private boolean IGNORE_DOMAINS_SPECIFIC_TO_ONE_SPECIES_OPTION_DEFAULT = false;
- final static private String MATRIX_MEAN_SCORE_BASED_GENOME_DISTANCE_SUFFIX = "_mean_score.pwd";
- final static private String MATRIX_SHARED_DOMAINS_BASED_GENOME_DISTANCE_SUFFIX = "_domains.pwd";
- final static private String MATRIX_SHARED_BIN_COMBINATIONS_BASED_GENOME_DISTANCE_SUFFIX = "_bin_combinations.pwd";
- final static private String NJ_TREE_MEAN_SCORE_BASED_GENOME_DISTANCE_SUFFIX = "_mean_score_NJ"
- + ForesterConstants.PHYLO_XML_SUFFIX;
- final static private String NJ_TREE_SHARED_DOMAINS_BASED_GENOME_DISTANCE_SUFFIX = "_domains_NJ"
- + ForesterConstants.PHYLO_XML_SUFFIX;
- final static private String NJ_TREE_SHARED_BIN_COMBINATIONS_BASED_GENOME_DISTANCE_SUFFIX = "_bin_combinations_NJ"
- + ForesterConstants.PHYLO_XML_SUFFIX;
- final static private String JACKNIFE_OPTION = "jack";
- final static private String JACKNIFE_RANDOM_SEED_OPTION = "seed";
- final static private String JACKNIFE_RATIO_OPTION = "jack_ratio";
- private static final int JACKNIFE_NUMBER_OF_RESAMPLINGS_DEFAULT = 100;
- final static private long JACKNIFE_RANDOM_SEED_DEFAULT = 19;
- final static private double JACKNIFE_RATIO_DEFAULT = 0.5;
+ public final static String PRG_NAME = "surfacing";
+ public static final String DOMAINS_PARSIMONY_TREE_OUTPUT_SUFFIX_DOLLO = "_d_dollo"
+ + ForesterConstants.PHYLO_XML_SUFFIX;
+ public static final String DOMAINS_PARSIMONY_TREE_OUTPUT_SUFFIX_FITCH = "_d_fitch"
+ + ForesterConstants.PHYLO_XML_SUFFIX;
+ public static final String BINARY_DOMAIN_COMBINATIONS_PARSIMONY_TREE_OUTPUT_SUFFIX_DOLLO = "_dc_dollo"
+ + ForesterConstants.PHYLO_XML_SUFFIX;
+ public static final String BINARY_DOMAIN_COMBINATIONS_PARSIMONY_TREE_OUTPUT_SUFFIX_FITCH = "_dc_fitch"
+ + ForesterConstants.PHYLO_XML_SUFFIX;
+ public static final String NEXUS_EXTERNAL_DOMAINS = "_dom.nex";
+ public static final String NEXUS_EXTERNAL_DOMAIN_COMBINATIONS = "_dc.nex";
+ public static final String NEXUS_SECONDARY_FEATURES = "_secondary_features.nex";
+ public static final String PARSIMONY_OUTPUT_GL_SUFFIX_DOLLO_SECONDARY_FEATURES = "_dollo_gl_secondary_features";
+ public static final String PARSIMONY_OUTPUT_GL_COUNTS_SUFFIX_DOLLO_SECONDARY_FEATURES = "_dollo_glc_secondary_features";
+ public static final String PARSIMONY_OUTPUT_DOLLO_GAINS_SECONDARY_FEATURES = "_dollo_gains_secondary_features";
+ public static final String PARSIMONY_OUTPUT_DOLLO_LOSSES_SECONDARY_FEATURES = "_dollo_losses_secondary_features";
+ public static final String PARSIMONY_OUTPUT_DOLLO_PRESENT_SECONDARY_FEATURES = "_dollo_present_secondary_features";
+ public static final String SECONDARY_FEATURES_PARSIMONY_TREE_OUTPUT_SUFFIX_DOLLO = "_secondary_features_dollo"
+ + ForesterConstants.PHYLO_XML_SUFFIX;
+ public static final String PARSIMONY_OUTPUT_DOLLO_ALL_GOID_D_ALL_NAMESPACES = "_dollo_goid_d";
+ public static final String PARSIMONY_OUTPUT_FITCH_ALL_GOID_BC_ALL_NAMESPACES = "_fitch_goid_dc";
+ final static private String HELP_OPTION_1 = "help";
+ final static private String HELP_OPTION_2 = "h";
+ final static private String OUTPUT_DIR_OPTION = "out_dir";
+ final static private String SCORING_OPTION = "scoring";
+ private static final DomainSimilarityScoring SCORING_DEFAULT = DomainSimilarity.DomainSimilarityScoring.COMBINATIONS;
+ final static private String SCORING_DOMAIN_COUNT_BASED = "domains";
+ final static private String SCORING_PROTEIN_COUNT_BASED = "proteins";
+ final static private String SCORING_COMBINATION_BASED = "combinations";
+ final static private String DETAILEDNESS_OPTION = "detail";
+ private final static Detailedness DETAILEDNESS_DEFAULT = DomainSimilarityCalculator.Detailedness.PUNCTILIOUS;
+ final static private String SPECIES_MATRIX_OPTION = "smatrix";
+ final static private String DETAILEDNESS_BASIC = "basic";
+ final static private String DETAILEDNESS_LIST_IDS = "list_ids";
+ final static private String DETAILEDNESS_PUNCTILIOUS = "punctilious";
+ final static private String DOMAIN_SIMILARITY_SORT_OPTION = "sort";
+ private static final DomainSimilaritySortField DOMAIN_SORT_FILD_DEFAULT = DomainSimilarity.DomainSimilaritySortField.DOMAIN_ID;
+ final static private String DOMAIN_SIMILARITY_SORT_MIN = "min";
+ final static private String DOMAIN_SIMILARITY_SORT_MAX = "max";
+ final static private String DOMAIN_SIMILARITY_SORT_SD = "sd";
+ final static private String DOMAIN_SIMILARITY_SORT_MEAN = "mean";
+ final static private String DOMAIN_SIMILARITY_SORT_DIFF = "diff";
+ final static private String DOMAIN_SIMILARITY_SORT_COUNTS_DIFF = "count_diff";
+ final static private String DOMAIN_SIMILARITY_SORT_ABS_COUNTS_DIFF = "abs_count_diff";
+ final static private String DOMAIN_SIMILARITY_SORT_SPECIES_COUNT = "species";
+ final static private String DOMAIN_SIMILARITY_SORT_ALPHA = "alpha";
+ final static private String DOMAIN_SIMILARITY_SORT_BY_SPECIES_COUNT_FIRST_OPTION = "species_first";
+ final static private String DOMAIN_COUNT_SORT_OPTION = "dc_sort";
+ private static final GenomeWideCombinableDomainsSortOrder DOMAINS_SORT_ORDER_DEFAULT = GenomeWideCombinableDomains.GenomeWideCombinableDomainsSortOrder.ALPHABETICAL_KEY_ID;
+ final static private String DOMAIN_COUNT_SORT_ALPHA = "alpha";
+ final static private String DOMAIN_COUNT_SORT_KEY_DOMAIN_COUNT = "dom";
+ final static private String DOMAIN_COUNT_SORT_KEY_DOMAIN_PROTEINS_COUNT = "prot";
+ final static private String DOMAIN_COUNT_SORT_COMBINATIONS_COUNT = "comb";
+ final static private String CUTOFF_SCORE_FILE_OPTION = "cos";
+ final static private String NOT_IGNORE_DUFS_OPTION = "dufs";
+ final static private String MAX_E_VALUE_OPTION = "e";
+ final static private String MAX_ALLOWED_OVERLAP_OPTION = "mo";
+ final static private String NO_ENGULFING_OVERLAP_OPTION = "no_eo";
+ final static private String IGNORE_COMBINATION_WITH_SAME_OPTION = "ignore_self_comb";
+ final static private String PAIRWISE_DOMAIN_COMPARISONS_PREFIX = "pwc_";
+ final static private String PAIRWISE_DOMAIN_COMPARISONS_OPTION = "pwc";
+ final static private String OUTPUT_FILE_OPTION = "o";
+ final static private String PFAM_TO_GO_FILE_USE_OPTION = "p2g";
+ final static private String GO_OBO_FILE_USE_OPTION = "obo";
+ final static private String GO_NAMESPACE_LIMIT_OPTION = "go_namespace";
+ final static private String GO_NAMESPACE_LIMIT_OPTION_MOLECULAR_FUNCTION = "molecular_function";
+ final static private String GO_NAMESPACE_LIMIT_OPTION_BIOLOGICAL_PROCESS = "biological_process";
+ final static private String GO_NAMESPACE_LIMIT_OPTION_CELLULAR_COMPONENT = "cellular_component";
+ final static private String SECONDARY_FEATURES_PARSIMONY_MAP_FILE = "secondary";
+ final static private String DOMAIN_SIMILARITY_PRINT_OPTION_SIMPLE_TAB_DELIMITED = "simple_tab";
+ final static private String DOMAIN_SIMILARITY_PRINT_OPTION_SIMPLE_HTML = "simple_html";
+ final static private String DOMAIN_SIMILARITY_PRINT_OPTION_DETAILED_HTML = "detailed_html";
+ final static private String DOMAIN_SIMILARITY_PRINT_OPTION = "ds_output";
+ private static final PRINT_OPTION DOMAIN_SIMILARITY_PRINT_OPTION_DEFAULT = PrintableDomainSimilarity.PRINT_OPTION.HTML;
+ final static private String IGNORE_DOMAINS_WITHOUT_COMBINATIONS_IN_ALL_SPECIES_OPTION = "ignore_singlet_domains";
+ final static private String IGNORE_VIRAL_IDS = "ignore_viral_ids";
+ final static private boolean IGNORE_DOMAINS_WITHOUT_COMBINATIONS_IN_ALL_SPECIES_DEFAULT = false;
+ final static private String IGNORE_DOMAINS_SPECIFIC_TO_ONE_SPECIES_OPTION = "ignore_species_specific_domains";
+ final static private boolean IGNORE_DOMAINS_SPECIFIC_TO_ONE_SPECIES_OPTION_DEFAULT = false;
+ final static private String MATRIX_MEAN_SCORE_BASED_GENOME_DISTANCE_SUFFIX = "_mean_score.pwd";
+ final static private String MATRIX_SHARED_DOMAINS_BASED_GENOME_DISTANCE_SUFFIX = "_domains.pwd";
+ final static private String MATRIX_SHARED_BIN_COMBINATIONS_BASED_GENOME_DISTANCE_SUFFIX = "_bin_combinations.pwd";
+ final static private String NJ_TREE_MEAN_SCORE_BASED_GENOME_DISTANCE_SUFFIX = "_mean_score_NJ"
+ + ForesterConstants.PHYLO_XML_SUFFIX;
+ final static private String NJ_TREE_SHARED_DOMAINS_BASED_GENOME_DISTANCE_SUFFIX = "_domains_NJ"
+ + ForesterConstants.PHYLO_XML_SUFFIX;
+ final static private String NJ_TREE_SHARED_BIN_COMBINATIONS_BASED_GENOME_DISTANCE_SUFFIX = "_bin_combinations_NJ"
+ + ForesterConstants.PHYLO_XML_SUFFIX;
+ final static private String JACKNIFE_OPTION = "jack";
+ final static private String JACKNIFE_RANDOM_SEED_OPTION = "seed";
+ final static private String JACKNIFE_RATIO_OPTION = "jack_ratio";
+ private static final int JACKNIFE_NUMBER_OF_RESAMPLINGS_DEFAULT = 100;
+ final static private long JACKNIFE_RANDOM_SEED_DEFAULT = 19;
+ final static private double JACKNIFE_RATIO_DEFAULT = 0.5;
//final static private String INFER_SPECIES_TREES_OPTION = "species_tree_inference";
- final static private String INFERRED_SD_BASED_NJ_SPECIES_TREE_SUFFIX = "_sd_nj.nh";
- final static private String INFERRED_SBC_BASED_NJ_SPECIES_TREE_SUFFIX = "_sbc_nj.nh";
- final static private String FILTER_POSITIVE_OPTION = "pos_filter";
- final static private String FILTER_NEGATIVE_OPTION = "neg_filter";
- final static private String FILTER_NEGATIVE_DOMAINS_OPTION = "neg_dom_filter";
- final static private String INPUT_FILES_FROM_FILE_OPTION = "input";
- final static private String INPUT_SPECIES_TREE_OPTION = "species_tree";
- final static private String SEQ_EXTRACT_OPTION = "prot_extract";
- final static private char SEPARATOR_FOR_INPUT_VALUES = '#';
- final static private String PRG_VERSION = "2.100";
- final static private String PRG_DATE = "2011.06.17";
- final static private String E_MAIL = "czmasek@burnham.org";
- final static private String WWW = "www.phylosoft.org/forester/applications/surfacing";
- final static private boolean IGNORE_DUFS_DEFAULT = true;
- final static private boolean IGNORE_COMBINATION_WITH_SAME_DEFAULLT = false;
- final static private double MAX_E_VALUE_DEFAULT = -1;
- final static private int MAX_ALLOWED_OVERLAP_DEFAULT = -1;
- private static final String RANDOM_SEED_FOR_FITCH_PARSIMONY_OPTION = "random_seed";
- private static final String CONSIDER_DOMAIN_COMBINATION_DIRECTEDNESS = "consider_bdc_direction";
- private static final String CONSIDER_DOMAIN_COMBINATION_DIRECTEDNESS_AND_ADJACENCY = "consider_bdc_adj";
- private static final String SEQ_EXTRACT_SUFFIX = ".prot";
- private static final String PLUS_MINUS_ANALYSIS_OPTION = "plus_minus";
- private static final String PLUS_MINUS_DOM_SUFFIX = "_plus_minus_dom.txt";
- private static final String PLUS_MINUS_DOM_SUFFIX_HTML = "_plus_minus_dom.html";
- private static final String PLUS_MINUS_DC_SUFFIX_HTML = "_plus_minus_dc.html";
- private static final int PLUS_MINUS_ANALYSIS_MIN_DIFF_DEFAULT = 0;
- private static final double PLUS_MINUS_ANALYSIS_FACTOR_DEFAULT = 1.0;
- private static final String PLUS_MINUS_ALL_GO_IDS_DOM_SUFFIX = "_plus_minus_go_ids_all.txt";
- private static final String PLUS_MINUS_PASSING_GO_IDS_DOM_SUFFIX = "_plus_minus_go_ids_passing.txt";
- private static final String OUTPUT_LIST_OF_ALL_PROTEINS_OPTIONS = "all_prot";
- private static final boolean VERBOSE = false;
- private static final String OUTPUT_DOMAIN_COMBINATIONS_GAINED_MORE_THAN_ONCE_ANALYSIS_SUFFIX = "_fitch_dc_gains_counts";
- private static final String OUTPUT_DOMAIN_COMBINATIONS_LOST_MORE_THAN_ONCE_ANALYSIS_SUFFIX = "_fitch_dc_losses_counts";
- private static final String DOMAIN_LENGTHS_ANALYSIS_SUFFIX = "_domain_lengths_analysis";
- private static final boolean PERFORM_DOMAIN_LENGTH_ANALYSIS = true;
- public static final String ALL_PFAMS_ENCOUNTERED_SUFFIX = "_all_encountered_pfams";
- public static final String ALL_PFAMS_ENCOUNTERED_WITH_GO_ANNOTATION_SUFFIX = "_all_encountered_pfams_with_go_annotation";
- public static final String ENCOUNTERED_PFAMS_SUMMARY_SUFFIX = "_encountered_pfams_summary";
- public static final String ALL_PFAMS_GAINED_AS_DOMAINS_SUFFIX = "_all_pfams_gained_as_domains";
- public static final String ALL_PFAMS_LOST_AS_DOMAINS_SUFFIX = "_all_pfams_lost_as_domains";
- public static final String ALL_PFAMS_GAINED_AS_DC_SUFFIX = "_all_pfams_gained_as_dc";
- public static final String ALL_PFAMS_LOST_AS_DC_SUFFIX = "_all_pfams_lost_as_dc";
- public static final String BASE_DIRECTORY_PER_NODE_DOMAIN_GAIN_LOSS_FILES = "PER_NODE_EVENTS";
- public static final String BASE_DIRECTORY_PER_SUBTREE_DOMAIN_GAIN_LOSS_FILES = "PER_SUBTREE_EVENTS";
- public static final String D_PROMISCUITY_FILE_SUFFIX = "_domain_promiscuities";
- private static final String LOG_FILE_SUFFIX = "_log.txt";
- private static final String DATA_FILE_SUFFIX = "_domain_combination_data.txt";
- private static final String DATA_FILE_DESC = "#SPECIES\tPRTEIN_ID\tN_TERM_DOMAIN\tC_TERM_DOMAIN\tN_TERM_DOMAIN_PER_DOMAIN_E_VALUE\tC_TERM_DOMAIN_PER_DOMAIN_E_VALUE\tN_TERM_DOMAIN_COUNTS_PER_PROTEIN\tC_TERM_DOMAIN_COUNTS_PER_PROTEIN";
- private static final INDIVIDUAL_SCORE_CUTOFF INDIVIDUAL_SCORE_CUTOFF_DEFAULT = INDIVIDUAL_SCORE_CUTOFF.FULL_SEQUENCE;
- public static final String INDEPENDENT_DC_GAINS_FITCH_PARS_COUNTS_OUTPUT_SUFFIX = "_indep_dc_gains_fitch_counts.txt";
- public static final String INDEPENDENT_DC_GAINS_FITCH_PARS_DC_OUTPUT_SUFFIX = "_indep_dc_gains_fitch_lists.txt";
- public static final String INDEPENDENT_DC_GAINS_FITCH_PARS_DC_FOR_GO_MAPPING_OUTPUT_SUFFIX = "_indep_dc_gains_fitch_lists_for_go_mapping.txt";
- public static final String INDEPENDENT_DC_GAINS_FITCH_PARS_DC_FOR_GO_MAPPING_OUTPUT_UNIQUE_SUFFIX = "_indep_dc_gains_fitch_lists_for_go_mapping_unique.txt";
+ final static private String INFERRED_SD_BASED_NJ_SPECIES_TREE_SUFFIX = "_sd_nj.nh";
+ final static private String INFERRED_SBC_BASED_NJ_SPECIES_TREE_SUFFIX = "_sbc_nj.nh";
+ final static private String FILTER_POSITIVE_OPTION = "pos_filter";
+ final static private String FILTER_NEGATIVE_OPTION = "neg_filter";
+ final static private String FILTER_NEGATIVE_DOMAINS_OPTION = "neg_dom_filter";
+ final static private String INPUT_FILES_FROM_FILE_OPTION = "input";
+ final static private String INPUT_SPECIES_TREE_OPTION = "species_tree";
+ final static private String SEQ_EXTRACT_OPTION = "prot_extract";
+ final static private char SEPARATOR_FOR_INPUT_VALUES = '#';
+ final static private String PRG_VERSION = "2.250";
+ final static private String PRG_DATE = "2012.05.07";
+ final static private String E_MAIL = "czmasek@burnham.org";
+ final static private String WWW = "www.phylosoft.org/forester/applications/surfacing";
+ final static private boolean IGNORE_DUFS_DEFAULT = true;
+ final static private boolean IGNORE_COMBINATION_WITH_SAME_DEFAULLT = false;
+ final static private double MAX_E_VALUE_DEFAULT = -1;
+ final static private int MAX_ALLOWED_OVERLAP_DEFAULT = -1;
+ private static final String RANDOM_SEED_FOR_FITCH_PARSIMONY_OPTION = "random_seed";
+ private static final String CONSIDER_DOMAIN_COMBINATION_DIRECTEDNESS = "consider_bdc_direction";
+ private static final String CONSIDER_DOMAIN_COMBINATION_DIRECTEDNESS_AND_ADJACENCY = "consider_bdc_adj";
+ private static final String SEQ_EXTRACT_SUFFIX = ".prot";
+ private static final String PLUS_MINUS_ANALYSIS_OPTION = "plus_minus";
+ private static final String PLUS_MINUS_DOM_SUFFIX = "_plus_minus_dom.txt";
+ private static final String PLUS_MINUS_DOM_SUFFIX_HTML = "_plus_minus_dom.html";
+ private static final String PLUS_MINUS_DC_SUFFIX_HTML = "_plus_minus_dc.html";
+ private static final int PLUS_MINUS_ANALYSIS_MIN_DIFF_DEFAULT = 0;
+ private static final double PLUS_MINUS_ANALYSIS_FACTOR_DEFAULT = 1.0;
+ private static final String PLUS_MINUS_ALL_GO_IDS_DOM_SUFFIX = "_plus_minus_go_ids_all.txt";
+ private static final String PLUS_MINUS_PASSING_GO_IDS_DOM_SUFFIX = "_plus_minus_go_ids_passing.txt";
+ private static final String OUTPUT_LIST_OF_ALL_PROTEINS_OPTIONS = "all_prot";
+ private static final boolean VERBOSE = false;
+ private static final String OUTPUT_DOMAIN_COMBINATIONS_GAINED_MORE_THAN_ONCE_ANALYSIS_SUFFIX = "_fitch_dc_gains_counts";
+ private static final String OUTPUT_DOMAIN_COMBINATIONS_LOST_MORE_THAN_ONCE_ANALYSIS_SUFFIX = "_fitch_dc_losses_counts";
+ private static final String DOMAIN_LENGTHS_ANALYSIS_SUFFIX = "_domain_lengths_analysis";
+ private static final boolean PERFORM_DOMAIN_LENGTH_ANALYSIS = true;
+ public static final String ALL_PFAMS_ENCOUNTERED_SUFFIX = "_all_encountered_pfams";
+ public static final String ALL_PFAMS_ENCOUNTERED_WITH_GO_ANNOTATION_SUFFIX = "_all_encountered_pfams_with_go_annotation";
+ public static final String ENCOUNTERED_PFAMS_SUMMARY_SUFFIX = "_encountered_pfams_summary";
+ public static final String ALL_PFAMS_GAINED_AS_DOMAINS_SUFFIX = "_all_pfams_gained_as_domains";
+ public static final String ALL_PFAMS_LOST_AS_DOMAINS_SUFFIX = "_all_pfams_lost_as_domains";
+ public static final String ALL_PFAMS_GAINED_AS_DC_SUFFIX = "_all_pfams_gained_as_dc";
+ public static final String ALL_PFAMS_LOST_AS_DC_SUFFIX = "_all_pfams_lost_as_dc";
+ public static final String BASE_DIRECTORY_PER_NODE_DOMAIN_GAIN_LOSS_FILES = "PER_NODE_EVENTS";
+ public static final String BASE_DIRECTORY_PER_SUBTREE_DOMAIN_GAIN_LOSS_FILES = "PER_SUBTREE_EVENTS";
+ public static final String D_PROMISCUITY_FILE_SUFFIX = "_domain_promiscuities";
+ private static final String LOG_FILE_SUFFIX = "_log.txt";
+ private static final String DATA_FILE_SUFFIX = "_domain_combination_data.txt";
+ private static final String DATA_FILE_DESC = "#SPECIES\tPRTEIN_ID\tN_TERM_DOMAIN\tC_TERM_DOMAIN\tN_TERM_DOMAIN_PER_DOMAIN_E_VALUE\tC_TERM_DOMAIN_PER_DOMAIN_E_VALUE\tN_TERM_DOMAIN_COUNTS_PER_PROTEIN\tC_TERM_DOMAIN_COUNTS_PER_PROTEIN";
+ private static final INDIVIDUAL_SCORE_CUTOFF INDIVIDUAL_SCORE_CUTOFF_DEFAULT = INDIVIDUAL_SCORE_CUTOFF.FULL_SEQUENCE;
+ public static final String INDEPENDENT_DC_GAINS_FITCH_PARS_COUNTS_OUTPUT_SUFFIX = "_indep_dc_gains_fitch_counts.txt";
+ public static final String INDEPENDENT_DC_GAINS_FITCH_PARS_DC_OUTPUT_SUFFIX = "_indep_dc_gains_fitch_lists.txt";
+ public static final String INDEPENDENT_DC_GAINS_FITCH_PARS_DC_FOR_GO_MAPPING_OUTPUT_SUFFIX = "_indep_dc_gains_fitch_lists_for_go_mapping.txt";
+ public static final String INDEPENDENT_DC_GAINS_FITCH_PARS_DC_FOR_GO_MAPPING_OUTPUT_UNIQUE_SUFFIX = "_indep_dc_gains_fitch_lists_for_go_mapping_unique.txt";
+ public static final String LIMIT_SPEC_FOR_PROT_EX = null; // e.g. "HUMAN"; set to null for not using this feature (default).
+ public static final String BINARY_DOMAIN_COMBINATIONS_PARSIMONY_TREE_OUTPUT_SUFFIX_FITCH_MAPPED = "_dc_MAPPED_secondary_features_fitch"
+ + ForesterConstants.PHYLO_XML_SUFFIX;
+ public static final String INDEPENDENT_DC_GAINS_FITCH_PARS_COUNTS_MAPPED_OUTPUT_SUFFIX = "_indep_dc_gains_fitch_counts_MAPPED.txt";
+ public static final String INDEPENDENT_DC_GAINS_FITCH_PARS_DC_MAPPED_OUTPUT_SUFFIX = "_indep_dc_gains_fitch_lists_MAPPED.txt";
+ public static final String INDEPENDENT_DC_GAINS_FITCH_PARS_DC_FOR_GO_MAPPING_MAPPED_OUTPUT_SUFFIX = "_indep_dc_gains_fitch_lists_for_go_mapping_MAPPED.txt";
+ public static final String INDEPENDENT_DC_GAINS_FITCH_PARS_DC_FOR_GO_MAPPING_MAPPED_OUTPUT_UNIQUE_SUFFIX = "_indep_dc_gains_fitch_lists_for_go_mapping_unique_MAPPED.txt";
+ private static final boolean PERFORM_DC_REGAIN_PROTEINS_STATS = true;
private static void checkWriteabilityForPairwiseComparisons( final PrintableDomainSimilarity.PRINT_OPTION domain_similarity_print_option,
final String[][] input_file_properties,
* @param sum_of_all_domains_encountered
* @param all_bin_domain_combinations_encountered
* @param is_gains_analysis
+ * @param protein_length_stats_by_dc
* @throws IOException
*/
private static void executeFitchGainsAnalysis( final File output_file,
}
try {
final Phylogeny[] p_array = ParserBasedPhylogenyFactory.getInstance()
- .create( intree_file, ForesterUtil.createParserDependingOnFileType( intree_file, true ) );
+ .create( intree_file, ParserUtils.createParserDependingOnFileType( intree_file, true ) );
if ( p_array.length < 1 ) {
ForesterUtil.fatalError( surfacing.PRG_NAME, "file [" + intree_file
+ "] does not contain any phylogeny in phyloXML format" );
return intrees;
}
- private static List<Phylogeny> inferSpeciesTrees( final File outfile, final List<DistanceMatrix> distances_list ) {
+ private static List<Phylogeny> inferSpeciesTrees( final File outfile,
+ final List<BasicSymmetricalDistanceMatrix> distances_list ) {
final NeighborJoining nj = NeighborJoining.createInstance();
final List<Phylogeny> phylogenies = nj.execute( distances_list );
final PhylogenyWriter w = new PhylogenyWriter();
System.out.println( "Ignore combination with self: " + ignore_combination_with_same );
html_desc.append( "<tr><td>Ignore combination with self for domain combination similarity analyses:</td><td>"
+ ignore_combination_with_same + "</td></tr>" + nl );
- ;
System.out.println( "Consider directedness : "
+ ( dc_type != BinaryDomainCombination.DomainCombinationType.BASIC ) );
html_desc.append( "<tr><td>Consider directedness of binary domain combinations:</td><td>"
}
} // if ( perform_pwc ) {
System.out.println();
- html_desc.append( "<tr><td>Command line:</td><td>" + cla.getCommandLineArgsAsString() + "</td></tr>" + nl );
+ html_desc.append( "<tr><td>Command line:</td><td>\n" + cla.getCommandLineArgsAsString() + "\n</td></tr>" + nl );
System.out.println( "Command line : " + cla.getCommandLineArgsAsString() );
BufferedWriter[] query_domains_writer_ary = null;
List<DomainId>[] query_domain_ids_array = null;
catch ( final IOException e2 ) {
ForesterUtil.fatalError( surfacing.PRG_NAME, e2.getMessage() );
}
+ final DescriptiveStatistics protein_coverage_stats = new BasicDescriptiveStatistics();
+ final DescriptiveStatistics all_genomes_domains_per_potein_stats = new BasicDescriptiveStatistics();
+ final SortedMap<Integer, Integer> all_genomes_domains_per_potein_histo = new TreeMap<Integer, Integer>();
+ final SortedSet<String> domains_which_are_always_single = new TreeSet<String>();
+ final SortedSet<String> domains_which_are_sometimes_single_sometimes_not = new TreeSet<String>();
+ final SortedSet<String> domains_which_never_single = new TreeSet<String>();
+ BufferedWriter domains_per_potein_stats_writer = null;
+ try {
+ domains_per_potein_stats_writer = new BufferedWriter( new FileWriter( out_dir + ForesterUtil.FILE_SEPARATOR
+ + output_file + "__domains_per_potein_stats.txt" ) );
+ domains_per_potein_stats_writer.write( "Genome" );
+ domains_per_potein_stats_writer.write( "\t" );
+ domains_per_potein_stats_writer.write( "Mean" );
+ domains_per_potein_stats_writer.write( "\t" );
+ domains_per_potein_stats_writer.write( "SD" );
+ domains_per_potein_stats_writer.write( "\t" );
+ domains_per_potein_stats_writer.write( "Median" );
+ domains_per_potein_stats_writer.write( "\t" );
+ domains_per_potein_stats_writer.write( "N" );
+ domains_per_potein_stats_writer.write( "\t" );
+ domains_per_potein_stats_writer.write( "Min" );
+ domains_per_potein_stats_writer.write( "\t" );
+ domains_per_potein_stats_writer.write( "Max" );
+ domains_per_potein_stats_writer.write( "\n" );
+ }
+ catch ( final IOException e3 ) {
+ e3.printStackTrace();
+ }
+ Map<String, DescriptiveStatistics> protein_length_stats_by_dc = null;
+ Map<String, DescriptiveStatistics> domain_number_stats_by_dc = null;
+ final Map<String, DescriptiveStatistics> domain_length_stats_by_domain = new HashMap<String, DescriptiveStatistics>();
+ if ( PERFORM_DC_REGAIN_PROTEINS_STATS ) {
+ protein_length_stats_by_dc = new HashMap<String, DescriptiveStatistics>();
+ domain_number_stats_by_dc = new HashMap<String, DescriptiveStatistics>();
+ }
+ // Main loop:
for( int i = 0; i < number_of_genomes; ++i ) {
System.out.println();
System.out.println( ( i + 1 ) + "/" + number_of_genomes );
input_file_properties[ i ][ 1 ],
filter,
filter_type,
- ind_score_cutoff );
+ ind_score_cutoff,
+ true );
}
else {
parser = new HmmscanPerDomainTableParser( new File( input_file_properties[ i ][ 0 ] ),
input_file_properties[ i ][ 1 ],
- ind_score_cutoff );
+ ind_score_cutoff,
+ true );
}
if ( e_value_max >= 0.0 ) {
parser.setEValueMaximum( e_value_max );
System.out.println( "Domains ignored due to virus like id: " );
ForesterUtil.printCountingMap( parser.getDomainsIgnoredDueToVirusLikeIdCountsMap() );
}
+ final double coverage = ( double ) protein_list.size() / parser.getProteinsEncountered();
+ protein_coverage_stats.addValue( coverage );
System.out.println( "Number of proteins encountered : " + parser.getProteinsEncountered() );
log( "Number of proteins encountered : " + parser.getProteinsEncountered(), log_writer );
System.out.println( "Number of proteins stored : " + protein_list.size() );
log( "Number of proteins stored : " + protein_list.size(), log_writer );
+ System.out.println( "Coverage : "
+ + ForesterUtil.roundToInt( 100.0 * coverage ) + "%" );
+ log( "Coverage : " + ForesterUtil.roundToInt( 100.0 * coverage )
+ + "%", log_writer );
System.out.println( "Domains encountered : " + parser.getDomainsEncountered() );
log( "Domains encountered : " + parser.getDomainsEncountered(), log_writer );
System.out.println( "Domains stored : " + parser.getDomainsStored() );
dc_data_writer.write( SurfacingUtil.proteinToDomainCombinations( protein, count + "", "\t" )
.toString() );
++count;
+ for( final Domain d : protein.getProteinDomains() ) {
+ final String d_str = d.getDomainId().toString();
+ if ( !domain_length_stats_by_domain.containsKey( d_str ) ) {
+ domain_length_stats_by_domain.put( d_str, new BasicDescriptiveStatistics() );
+ }
+ domain_length_stats_by_domain.get( d_str ).addValue( d.getLength() );
+ }
}
}
catch ( final IOException e ) {
ForesterUtil.fatalError( surfacing.PRG_NAME, e.toString() );
}
+ SurfacingUtil.domainsPerProteinsStatistics( input_file_properties[ i ][ 1 ],
+ protein_list,
+ all_genomes_domains_per_potein_stats,
+ all_genomes_domains_per_potein_histo,
+ domains_which_are_always_single,
+ domains_which_are_sometimes_single_sometimes_not,
+ domains_which_never_single,
+ domains_per_potein_stats_writer );
gwcd_list.add( BasicGenomeWideCombinableDomains
.createInstance( protein_list,
ignore_combination_with_same,
new BasicSpecies( input_file_properties[ i ][ 1 ] ),
domain_id_to_go_ids_map,
- dc_type ) );
+ dc_type,
+ protein_length_stats_by_dc,
+ domain_number_stats_by_dc ) );
domain_lengths_table.addLengths( protein_list );
if ( gwcd_list.get( i ).getSize() > 0 ) {
SurfacingUtil.writeDomainCombinationsCountsFile( input_file_properties,
SurfacingUtil.extractProteinNames( protein_list,
query_domain_ids_array[ j ],
query_domains_writer_ary[ j ],
- "\t" );
+ "\t",
+ LIMIT_SPEC_FOR_PROT_EX );
query_domains_writer_ary[ j ].flush();
}
catch ( final IOException e ) {
}
System.gc();
} // for( int i = 0; i < number_of_genomes; ++i ) {
+ ForesterUtil.programMessage( PRG_NAME, "Wrote domain promiscuities to: "
+ + per_genome_domain_promiscuity_statistics_file );
+ //
try {
- per_genome_domain_promiscuity_statistics_writer.flush();
- per_genome_domain_promiscuity_statistics_writer.close();
- dc_data_writer.flush();
- dc_data_writer.close();
- log_writer.flush();
- log_writer.close();
+ domains_per_potein_stats_writer.write( "ALL" );
+ domains_per_potein_stats_writer.write( "\t" );
+ domains_per_potein_stats_writer.write( all_genomes_domains_per_potein_stats.arithmeticMean() + "" );
+ domains_per_potein_stats_writer.write( "\t" );
+ domains_per_potein_stats_writer.write( all_genomes_domains_per_potein_stats.sampleStandardDeviation() + "" );
+ domains_per_potein_stats_writer.write( "\t" );
+ domains_per_potein_stats_writer.write( all_genomes_domains_per_potein_stats.median() + "" );
+ domains_per_potein_stats_writer.write( "\t" );
+ domains_per_potein_stats_writer.write( all_genomes_domains_per_potein_stats.getN() + "" );
+ domains_per_potein_stats_writer.write( "\t" );
+ domains_per_potein_stats_writer.write( all_genomes_domains_per_potein_stats.getMin() + "" );
+ domains_per_potein_stats_writer.write( "\t" );
+ domains_per_potein_stats_writer.write( all_genomes_domains_per_potein_stats.getMax() + "" );
+ domains_per_potein_stats_writer.write( "\n" );
+ domains_per_potein_stats_writer.close();
+ printOutPercentageOfMultidomainProteins( all_genomes_domains_per_potein_histo, log_writer );
+ ForesterUtil.map2file( new File( out_dir + ForesterUtil.FILE_SEPARATOR + output_file
+ + "__all_genomes_domains_per_potein_histo.txt" ), all_genomes_domains_per_potein_histo, "\t", "\n" );
+ ForesterUtil.collection2file( new File( out_dir + ForesterUtil.FILE_SEPARATOR + output_file
+ + "__domains_always_single_.txt" ), domains_which_are_always_single, "\n" );
+ ForesterUtil.collection2file( new File( out_dir + ForesterUtil.FILE_SEPARATOR + output_file
+ + "__domains_single_or_combined.txt" ), domains_which_are_sometimes_single_sometimes_not, "\n" );
+ ForesterUtil.collection2file( new File( out_dir + ForesterUtil.FILE_SEPARATOR + output_file
+ + "__domains_always_combined.txt" ), domains_which_never_single, "\n" );
+ ForesterUtil.programMessage( PRG_NAME,
+ "Average of proteins with a least one domain assigned: "
+ + ( 100 * protein_coverage_stats.arithmeticMean() ) + "% (+/-"
+ + ( 100 * protein_coverage_stats.sampleStandardDeviation() ) + "%)" );
+ ForesterUtil.programMessage( PRG_NAME, "Range of proteins with a least one domain assigned: " + 100
+ * protein_coverage_stats.getMin() + "%-" + 100 * protein_coverage_stats.getMax() + "%" );
+ log( "Average of prot with a least one dom assigned : " + ( 100 * protein_coverage_stats.arithmeticMean() )
+ + "% (+/-" + ( 100 * protein_coverage_stats.sampleStandardDeviation() ) + "%)", log_writer );
+ log( "Range of prot with a least one dom assigned : " + 100 * protein_coverage_stats.getMin() + "%-"
+ + 100 * protein_coverage_stats.getMax() + "%", log_writer );
}
catch ( final IOException e2 ) {
ForesterUtil.fatalError( surfacing.PRG_NAME, e2.getLocalizedMessage() );
}
- ForesterUtil.programMessage( PRG_NAME, "Wrote domain promiscuities to: "
- + per_genome_domain_promiscuity_statistics_file );
if ( query_domains_writer_ary != null ) {
for( int j = 0; j < query_domain_ids_array.length; j++ ) {
try {
}
}
}
+ try {
+ per_genome_domain_promiscuity_statistics_writer.close();
+ dc_data_writer.close();
+ log_writer.close();
+ }
+ catch ( final IOException e2 ) {
+ ForesterUtil.fatalError( surfacing.PRG_NAME, e2.getLocalizedMessage() );
+ }
if ( PERFORM_DOMAIN_LENGTH_ANALYSIS ) {
try {
SurfacingUtil.executeDomainLengthAnalysis( input_file_properties,
DescriptiveStatistics pw_stats = null;
try {
String my_outfile = output_file.toString();
- if ( !my_outfile.endsWith( ".html" ) ) {
+ Map<Character, Writer> split_writers = null;
+ Writer writer = null;
+ if ( similarities.size() > MINIMAL_NUMBER_OF_SIMILARITIES_FOR_SPLITTING ) {
+ if ( my_outfile.endsWith( ".html" ) ) {
+ my_outfile = my_outfile.substring( 0, my_outfile.length() - 5 );
+ }
+ split_writers = new HashMap<Character, Writer>();
+ createSplitWriters( out_dir, my_outfile, split_writers );
+ }
+ else if ( !my_outfile.endsWith( ".html" ) ) {
my_outfile += ".html";
+ writer = new BufferedWriter( new FileWriter( out_dir + ForesterUtil.FILE_SEPARATOR + my_outfile ) );
}
- final Writer writer = new BufferedWriter( new FileWriter( out_dir == null ? my_outfile : out_dir
- + ForesterUtil.FILE_SEPARATOR + my_outfile ) );
List<Species> species_order = null;
if ( species_matrix ) {
species_order = new ArrayList<Species>();
.writeDomainSimilaritiesToFile( html_desc,
new StringBuilder( number_of_genomes + " genomes" ),
writer,
+ split_writers,
similarities,
number_of_genomes == 2,
species_order,
output_binary_domain_combinationsfor_graph_analysis,
all_bin_domain_combinations_gained_fitch,
all_bin_domain_combinations_lost_fitch,
- dc_type );
+ dc_type,
+ protein_length_stats_by_dc,
+ domain_number_stats_by_dc,
+ domain_length_stats_by_domain );
// Listing of all domain combinations gained is only done if only one input tree is used.
if ( ( domain_id_to_secondary_features_maps != null )
&& ( domain_id_to_secondary_features_maps.length > 0 ) ) {
final Runtime rt = java.lang.Runtime.getRuntime();
final long free_memory = rt.freeMemory() / 1000000;
final long total_memory = rt.totalMemory() / 1000000;
- System.out.println();
- System.out.println( "Time for analysis : " + ( new Date().getTime() - analysis_start_time ) + "ms" );
- System.out.println( "Total running time: " + ( new Date().getTime() - start_time ) + "ms " );
- System.out.println( "Free memory : " + free_memory + "MB, total memory: " + total_memory + "MB" );
- System.out.println();
- System.out.println( "If this application is useful to you, please cite:" );
- System.out.println( surfacing.WWW );
- System.out.println();
+ ForesterUtil.programMessage( PRG_NAME, "Time for analysis : " + ( new Date().getTime() - analysis_start_time )
+ + "ms" );
+ ForesterUtil.programMessage( PRG_NAME, "Total running time: " + ( new Date().getTime() - start_time ) + "ms " );
+ ForesterUtil.programMessage( PRG_NAME, "Free memory : " + free_memory + "MB, total memory: "
+ + total_memory + "MB" );
+ ForesterUtil.programMessage( PRG_NAME, "If this application is useful to you, please cite:" );
+ ForesterUtil.programMessage( PRG_NAME, surfacing.WWW );
ForesterUtil.programMessage( PRG_NAME, "OK" );
System.out.println();
}
+ private static void createSplitWriters( final File out_dir,
+ final String my_outfile,
+ final Map<Character, Writer> split_writers ) throws IOException {
+ split_writers.put( 'a', new BufferedWriter( new FileWriter( out_dir + ForesterUtil.FILE_SEPARATOR + my_outfile
+ + "_domains_A.html" ) ) );
+ split_writers.put( 'b', new BufferedWriter( new FileWriter( out_dir + ForesterUtil.FILE_SEPARATOR + my_outfile
+ + "_domains_B.html" ) ) );
+ split_writers.put( 'c', new BufferedWriter( new FileWriter( out_dir + ForesterUtil.FILE_SEPARATOR + my_outfile
+ + "_domains_C.html" ) ) );
+ split_writers.put( 'd', new BufferedWriter( new FileWriter( out_dir + ForesterUtil.FILE_SEPARATOR + my_outfile
+ + "_domains_D.html" ) ) );
+ split_writers.put( 'e', new BufferedWriter( new FileWriter( out_dir + ForesterUtil.FILE_SEPARATOR + my_outfile
+ + "_domains_E.html" ) ) );
+ split_writers.put( 'f', new BufferedWriter( new FileWriter( out_dir + ForesterUtil.FILE_SEPARATOR + my_outfile
+ + "_domains_F.html" ) ) );
+ split_writers.put( 'g', new BufferedWriter( new FileWriter( out_dir + ForesterUtil.FILE_SEPARATOR + my_outfile
+ + "_domains_G.html" ) ) );
+ split_writers.put( 'h', new BufferedWriter( new FileWriter( out_dir + ForesterUtil.FILE_SEPARATOR + my_outfile
+ + "_domains_H.html" ) ) );
+ split_writers.put( 'i', new BufferedWriter( new FileWriter( out_dir + ForesterUtil.FILE_SEPARATOR + my_outfile
+ + "_domains_I.html" ) ) );
+ split_writers.put( 'j', new BufferedWriter( new FileWriter( out_dir + ForesterUtil.FILE_SEPARATOR + my_outfile
+ + "_domains_J.html" ) ) );
+ split_writers.put( 'k', new BufferedWriter( new FileWriter( out_dir + ForesterUtil.FILE_SEPARATOR + my_outfile
+ + "_domains_K.html" ) ) );
+ split_writers.put( 'l', new BufferedWriter( new FileWriter( out_dir + ForesterUtil.FILE_SEPARATOR + my_outfile
+ + "_domains_L.html" ) ) );
+ split_writers.put( 'm', new BufferedWriter( new FileWriter( out_dir + ForesterUtil.FILE_SEPARATOR + my_outfile
+ + "_domains_M.html" ) ) );
+ split_writers.put( 'n', new BufferedWriter( new FileWriter( out_dir + ForesterUtil.FILE_SEPARATOR + my_outfile
+ + "_domains_N.html" ) ) );
+ split_writers.put( 'o', new BufferedWriter( new FileWriter( out_dir + ForesterUtil.FILE_SEPARATOR + my_outfile
+ + "_domains_O.html" ) ) );
+ split_writers.put( 'p', new BufferedWriter( new FileWriter( out_dir + ForesterUtil.FILE_SEPARATOR + my_outfile
+ + "_domains_P.html" ) ) );
+ split_writers.put( 'q', new BufferedWriter( new FileWriter( out_dir + ForesterUtil.FILE_SEPARATOR + my_outfile
+ + "_domains_Q.html" ) ) );
+ split_writers.put( 'r', new BufferedWriter( new FileWriter( out_dir + ForesterUtil.FILE_SEPARATOR + my_outfile
+ + "_domains_R.html" ) ) );
+ split_writers.put( 's', new BufferedWriter( new FileWriter( out_dir + ForesterUtil.FILE_SEPARATOR + my_outfile
+ + "_domains_S.html" ) ) );
+ split_writers.put( 't', new BufferedWriter( new FileWriter( out_dir + ForesterUtil.FILE_SEPARATOR + my_outfile
+ + "_domains_T.html" ) ) );
+ split_writers.put( 'u', new BufferedWriter( new FileWriter( out_dir + ForesterUtil.FILE_SEPARATOR + my_outfile
+ + "_domains_U.html" ) ) );
+ split_writers.put( 'v', new BufferedWriter( new FileWriter( out_dir + ForesterUtil.FILE_SEPARATOR + my_outfile
+ + "_domains_V.html" ) ) );
+ split_writers.put( 'w', new BufferedWriter( new FileWriter( out_dir + ForesterUtil.FILE_SEPARATOR + my_outfile
+ + "_domains_W.html" ) ) );
+ split_writers.put( 'x', new BufferedWriter( new FileWriter( out_dir + ForesterUtil.FILE_SEPARATOR + my_outfile
+ + "_domains_X.html" ) ) );
+ split_writers.put( 'y', new BufferedWriter( new FileWriter( out_dir + ForesterUtil.FILE_SEPARATOR + my_outfile
+ + "_domains_Y.html" ) ) );
+ split_writers.put( 'z', new BufferedWriter( new FileWriter( out_dir + ForesterUtil.FILE_SEPARATOR + my_outfile
+ + "_domains_Z.html" ) ) );
+ split_writers.put( '0', new BufferedWriter( new FileWriter( out_dir + ForesterUtil.FILE_SEPARATOR + my_outfile
+ + "_domains_0.html" ) ) );
+ }
+
+ private static void printOutPercentageOfMultidomainProteins( final SortedMap<Integer, Integer> all_genomes_domains_per_potein_histo,
+ final Writer log_writer ) {
+ int sum = 0;
+ for( final Entry<Integer, Integer> entry : all_genomes_domains_per_potein_histo.entrySet() ) {
+ sum += entry.getValue();
+ }
+ final double percentage = 100.0 * ( sum - all_genomes_domains_per_potein_histo.get( 1 ) ) / sum;
+ ForesterUtil.programMessage( PRG_NAME, "Percentage of multidomain proteins: " + percentage + "%" );
+ log( "Percentage of multidomain proteins: : " + percentage + "%", log_writer );
+ }
+
private static void preparePhylogenyForParsimonyAnalyses( final Phylogeny intree,
final String[][] input_file_properties ) {
final String[] genomes = new String[ input_file_properties.length ];
final PhylogenyNode n = it.next();
if ( ForesterUtil.isEmpty( n.getName() ) ) {
if ( n.getNodeData().isHasTaxonomy()
+ && !ForesterUtil.isEmpty( n.getNodeData().getTaxonomy().getTaxonomyCode() ) ) {
+ n.setName( n.getNodeData().getTaxonomy().getTaxonomyCode() );
+ }
+ else if ( n.getNodeData().isHasTaxonomy()
&& !ForesterUtil.isEmpty( n.getNodeData().getTaxonomy().getScientificName() ) ) {
n.setName( n.getNodeData().getTaxonomy().getScientificName() );
}
+ else if ( n.getNodeData().isHasTaxonomy()
+ && !ForesterUtil.isEmpty( n.getNodeData().getTaxonomy().getCommonName() ) ) {
+ n.setName( n.getNodeData().getTaxonomy().getCommonName() );
+ }
else {
- ForesterUtil.fatalError( surfacing.PRG_NAME,
- "node without both name and scientific taxonomy name found" );
+ ForesterUtil
+ .fatalError( surfacing.PRG_NAME,
+ "node with no name, scientific name, common name, or taxonomy code present" );
}
}
}
SurfacingUtil.checkForOutputFileWriteability( out );
try {
final Writer proteins_file_writer = new BufferedWriter( new FileWriter( out ) );
- SurfacingUtil.extractProteinNames( protein_lists_per_species, domain, proteins_file_writer, "\t" );
+ SurfacingUtil.extractProteinNames( protein_lists_per_species,
+ domain,
+ proteins_file_writer,
+ "\t",
+ LIMIT_SPEC_FOR_PROT_EX );
proteins_file_writer.close();
}
catch ( final IOException e ) {
git://source.jalview.org / jalview.git / blobdiff