de.julielab.geneexpbase.data.DocumentSourceFiles Maven / Gradle / Ivy

Go to download
package de.julielab.geneexpbase.data;

import de.julielab.geneexpbase.genemodel.GeneMention;

import java.io.File;
import java.util.List;
import java.util.Set;

public class DocumentSourceFiles {
    private String name;

    private String basePath;
    private String predictedGenesPath;
    private String sentencesPath;
    private String chunksPath;
    private String posPath;
    private String speciesPath;
    private String acronymsPath;
    private String corefPath;
    private String appositionsPath;
    private String meshPath;
    private String substancesPath;
    private String ontologyMentionsPath;
    private String docTextPath;
    private String goldGeneList;
    private List allowedGeneTypes;
    private Set taggersToUse;
    private boolean filterSpecies;
    private boolean addReferenceSpecies;
    private boolean hasGeneIds = true;
    private final GeneMention.SpecificType defaultEntityType = GeneMention.SpecificType.GENE;
    private boolean speciesCorpus;
    private boolean inferDocumentLevelLabelsToMentions;
    private boolean completelyAnnotated;
    private String nonGenePhrasesPath;
    private String gazetteerMatchesWithIds;

    public DocumentSourceFiles() {
        this.basePath = "";
    }

    public String getAppositionsPath() {
        return appositionsPath;
    }

    public void setAppositionsPath(String appositionsPath) {
        this.appositionsPath = resolveToBasePath(appositionsPath);
    }

    public String getCorefPath() {
        return corefPath;
    }

    public void setCorefPath(String corefPath) {
        this.corefPath = resolveToBasePath(corefPath);
    }


    public GeneMention.SpecificType getDefaultEntityType() {
        return defaultEntityType;
    }

    /**
     * @deprecated This setting should be set by a parameter
     */
    @Deprecated
    public boolean isAddReferenceSpecies() {
        return addReferenceSpecies;
    }

    /**
     * @deprecated This setting should be set by a parameter
     */
    @Deprecated
    public void setAddReferenceSpecies(boolean addReferenceSpecies) {
        this.addReferenceSpecies = addReferenceSpecies;
    }

    /**
     * @deprecated This setting should be set by a parameter
     */
    @Deprecated
    public boolean isFilterSpecies() {
        return filterSpecies;
    }

    /**
     * @deprecated This setting should be set by a parameter
     */
    @Deprecated
    public void setFilterSpecies(boolean filterSpecies) {
        this.filterSpecies = filterSpecies;
    }

    /**
     * A list of strings that match the "entity type" field of the input files. Only entity types listed here
     * are accepted as gene mentions.
     *
     * @return
     */
    public List getAllowedGeneTypes() {
        return allowedGeneTypes;
    }

    public void setAllowedGeneTypes(List allowedGeneTypes) {
        this.allowedGeneTypes = allowedGeneTypes;
    }

    public Set getTaggersToUse() {
        return taggersToUse;
    }

    public void setTaggersToUse(Set taggersToUse) {
        this.taggersToUse = taggersToUse;
    }

    public String getPosPath() {
        return posPath;
    }

    public void setPosPath(String posPath) {
        this.posPath = resolveToBasePath(posPath);
    }

    public String getBasePath() {
        return basePath;
    }

    /**
     * Sets a base path to which all other paths will be relative. Must be set before other paths are set that should be relative to the base path.
     * This is optional and can safely be omitted.
     *
     * @param basePath
     */
    public void setBasePath(String basePath) {
        this.basePath = basePath;
    }

    public String getPredictedGenesPath() {
        return predictedGenesPath;
    }

    public void setPredictedGenesPath(String predictedGenesPath) {
        this.predictedGenesPath = resolveToBasePath(predictedGenesPath);
    }

    public String getSentencesPath() {
        return sentencesPath;
    }

    public void setSentencesPath(String sentencesPath) {
        this.sentencesPath = resolveToBasePath(sentencesPath);
    }

    public String getChunksPath() {
        return chunksPath;
    }

    public void setChunksPath(String chunksPath) {
        this.chunksPath = resolveToBasePath(chunksPath);
    }

    public String getSpeciesPath() {
        return speciesPath;
    }

    public void setSpeciesPath(String speciesPath) {
        this.speciesPath = resolveToBasePath(speciesPath);
    }

    public String getAcronymsPath() {
        return acronymsPath;
    }

    public void setAcronymsPath(String acronymsPath) {
        this.acronymsPath = resolveToBasePath(acronymsPath);
    }

    public String getMeshPath() {
        return meshPath;
    }

    public void setMeshPath(String meshPath) {
        setMeshPath(meshPath, true);
    }

    public String getSubstancesPath() {
        return substancesPath;
    }

    public void setSubstancesPath(String substancesPath) {
        setSubstancesPath(substancesPath, true);
    }

    public String getDocTextPath() {
        return docTextPath;
    }

    public void setDocTextPath(String docTextPath) {
        this.docTextPath = resolveToBasePath(docTextPath);
    }

    public String getGoldGeneList() {
        return goldGeneList;
    }

    public void setGoldGeneList(String goldGeneList) {
        this.goldGeneList = resolveToBasePath(goldGeneList);
    }

    public void setMeshPath(String meshPath, boolean appendToBasePath) {
        this.meshPath = appendToBasePath ? resolveToBasePath(meshPath) : meshPath;
    }

    public void setSubstancesPath(String substancesPath, boolean appendtoBasePath) {
        this.substancesPath = appendtoBasePath ? resolveToBasePath(substancesPath) : substancesPath;
    }

    public boolean hashMesh() {
        return this.meshPath != null && !this.meshPath.isBlank();
    }

    public boolean hasSubstances() {
        return this.substancesPath != null && !this.substancesPath.isBlank();
    }

    private String resolveToBasePath(String path) {
        return this.basePath.isBlank() ? path : this.basePath + File.separator + path;
    }

    public String getName() {
        return name;
    }

    public void setName(String name) {
        this.name = name;
    }

    public boolean isHasGeneIds() {
        return hasGeneIds;
    }

    public void setHasGeneIds(boolean hasGeneIds) {
        this.hasGeneIds = hasGeneIds;
    }

    public boolean isSpeciesCorpus() {
        return speciesCorpus;
    }

    public void setSpeciesCorpus(boolean speciesCorpus) {
        this.speciesCorpus = speciesCorpus;
    }

    public boolean getInferDocumentLevelLabelsToMentions() {
        return inferDocumentLevelLabelsToMentions;
    }

    /**
     * If set to true, the {@link DocumentLoader} will make a best effort to derive gold mentions for corpora
     * that have only been annotated on the document level.
     * This is done by searching for candidates for each predicted gene mention and use the best scored candidate
     * that has a ID found in the set of valid IDs annotated for the document.
     *
     * @param inferDocumentLevelLabelsToMentions Whether to perform document annotation to mention inference.
     */
    public void setInferDocumentLevelLabelsToMentions(boolean inferDocumentLevelLabelsToMentions) {
        this.inferDocumentLevelLabelsToMentions = inferDocumentLevelLabelsToMentions;
    }

    /**
     * Some gene corpora have annotated all gene occurrences while other focus on the most important genes
     * with regard to a specific task.
     *
     * @return Whether all genes in this corpus have been annotated or only a subset.
     */
    public boolean isCompletelyAnnotated() {
        return completelyAnnotated;
    }

    /**
     * Some gene corpora have annotated all gene occurrences while other focus on the most important genes
     * with regard to a specific task.
     *
     * @param completelyAnnotated  Whether all genes in this corpus have been annotated or only a subset.
     */
    public void setCompletelyAnnotated(boolean completelyAnnotated) {

        this.completelyAnnotated = completelyAnnotated;
    }

    public String getOntologyMentionsPath() {
        return ontologyMentionsPath;
    }

    public void setOntologyMentionsPath(String ontologyMentionsPath) {
        this.ontologyMentionsPath = resolveToBasePath(ontologyMentionsPath);
    }

    public void setNonGenePhrasesPath(String nonGenePhrasesPath) {
        this.nonGenePhrasesPath = resolveToBasePath(nonGenePhrasesPath);
    }

    public String getNonGenePhrasesPath() {
        return nonGenePhrasesPath;
    }

    public String getGazetteerMatchesWithIds() {
        return gazetteerMatchesWithIds;
    }

    public void setGazetteerMatchesWithIds(String gazetteerMatchesWithIds) {
        this.gazetteerMatchesWithIds = resolveToBasePath(gazetteerMatchesWithIds);
    }
}