Examples with ProteinSequenceEncoder edu.sdsc.mmtf.spark.ml.ProteinSequenceEncoder used on opensource projects

Example 6 with ProteinSequenceEncoder

use of edu.sdsc.mmtf.spark.ml.ProteinSequenceEncoder in project mmtf-spark by sbl-sdsc.

the class SecondaryStructureBlosum62Encoder method main.

/**
 * @param args args[0] outputFilePath, args[1] outputFormat (json|parquet)
 * @throws IOException
 * @throws StructureException
 */
public static void main(String[] args) throws IOException {
    String path = MmtfReader.getMmtfReducedPath();
    if (args.length != 2) {
        System.err.println("Usage: " + SecondaryStructureBlosum62Encoder.class.getSimpleName() + " <outputFilePath> + <fileFormat>");
        System.exit(1);
    }
    long start = System.nanoTime();
    SparkConf conf = new SparkConf().setMaster("local[*]").setAppName(SecondaryStructureBlosum62Encoder.class.getSimpleName());
    JavaSparkContext sc = new JavaSparkContext(conf);
    // read MMTF Hadoop sequence file and create a non-redundant Pisces
    // subset set (<=20% seq. identity) of L-protein chains
    int sequenceIdentity = 20;
    double resolution = 3.0;
    JavaPairRDD<String, StructureDataInterface> pdb = MmtfReader.readSequenceFile(path, sc).flatMapToPair(new StructureToPolymerChains()).filter(new Pisces(sequenceIdentity, resolution));
    int segmentLength = 11;
    Dataset<Row> data = SecondaryStructureSegmentExtractor.getDataset(pdb, segmentLength).cache();
    System.out.println("original data     : " + data.count());
    data = data.dropDuplicates("labelQ3", "sequence").cache();
    System.out.println("- duplicate Q3/seq: " + data.count());
    data = data.dropDuplicates("sequence").cache();
    System.out.println("- duplicate seq   : " + data.count());
    // add a property encoded feature vector
    ProteinSequenceEncoder encoder = new ProteinSequenceEncoder(data);
    data = encoder.blosum62Encode();
    data.printSchema();
    data.show(25, false);
    if (args[1].equals("json")) {
        // coalesce data into a single file
        data = data.coalesce(1);
    }
    data.write().mode("overwrite").format(args[1]).save(args[0]);
    long end = System.nanoTime();
    System.out.println(TimeUnit.NANOSECONDS.toSeconds(end - start) + " sec.");
}

Example 7 with ProteinSequenceEncoder

use of edu.sdsc.mmtf.spark.ml.ProteinSequenceEncoder in project mmtf-spark by sbl-sdsc.

the class SecondaryStructureOneHotEncoder method main.

/**
 * @param args args[0] outputFilePath, args[1] outputFormat (json|parquet)
 * @throws IOException
 * @throws StructureException
 */
public static void main(String[] args) throws IOException {
    String path = MmtfReader.getMmtfReducedPath();
    if (args.length < 2) {
        System.err.println("Usage: " + SecondaryStructureOneHotEncoder.class.getSimpleName() + " <outputFilePath> + <fileFormat> + [<modelFileName>]");
        System.exit(1);
    }
    long start = System.nanoTime();
    SparkConf conf = new SparkConf().setMaster("local[*]").setAppName(SecondaryStructureOneHotEncoder.class.getSimpleName());
    JavaSparkContext sc = new JavaSparkContext(conf);
    // read MMTF Hadoop sequence file and create a non-redundant Pisces
    // subset set (<=20% seq. identity) of L-protein chains
    int sequenceIdentity = 20;
    double resolution = 3.0;
    JavaPairRDD<String, StructureDataInterface> pdb = MmtfReader.readSequenceFile(path, sc).flatMapToPair(new StructureToPolymerChains()).filter(new Pisces(sequenceIdentity, resolution));
    // get content
    int segmentLength = 11;
    Dataset<Row> data = SecondaryStructureSegmentExtractor.getDataset(pdb, segmentLength).cache();
    System.out.println("original data     : " + data.count());
    data = data.dropDuplicates("labelQ3", "sequence").cache();
    System.out.println("- duplicate Q3/seq: " + data.count());
    data = data.dropDuplicates("sequence").cache();
    System.out.println("- duplicate seq   : " + data.count());
    // add one-hot encoded sequence feature vector to dataset
    ProteinSequenceEncoder encoder = new ProteinSequenceEncoder(data);
    data = encoder.oneHotEncode();
    data.printSchema();
    data.show(25, false);
    if (args[1].equals("json")) {
        // coalesce data into a single file
        data = data.coalesce(1);
    }
    data.write().mode("overwrite").format(args[1]).save(args[0]);
    long end = System.nanoTime();
    System.out.println(TimeUnit.NANOSECONDS.toSeconds(end - start) + " sec.");
}

Example 8 with ProteinSequenceEncoder

use of edu.sdsc.mmtf.spark.ml.ProteinSequenceEncoder in project mmtf-spark by sbl-sdsc.

the class SecondaryStructurePropertyEncoder method main.

/**
 * @param args outputFilePath outputFormat (json|parquet)
 * @throws IOException
 * @throws StructureException
 */
public static void main(String[] args) throws IOException {
    String path = MmtfReader.getMmtfReducedPath();
    if (args.length != 2) {
        System.err.println("Usage: " + SecondaryStructurePropertyEncoder.class.getSimpleName() + " <outputFilePath> + <fileFormat>");
        System.exit(1);
    }
    long start = System.nanoTime();
    SparkConf conf = new SparkConf().setMaster("local[*]").setAppName(SecondaryStructurePropertyEncoder.class.getSimpleName());
    JavaSparkContext sc = new JavaSparkContext(conf);
    // read MMTF Hadoop sequence file and create a non-redundant Pisces
    // subset set (<=20% seq. identity) of L-protein chains
    int sequenceIdentity = 20;
    double resolution = 3.0;
    JavaPairRDD<String, StructureDataInterface> pdb = MmtfReader.readSequenceFile(path, sc).flatMapToPair(new StructureToPolymerChains()).filter(new Pisces(sequenceIdentity, resolution));
    // get content
    int segmentLength = 11;
    Dataset<Row> data = SecondaryStructureSegmentExtractor.getDataset(pdb, segmentLength).cache();
    System.out.println("original data     : " + data.count());
    data = data.dropDuplicates("labelQ3", "sequence").cache();
    System.out.println("- duplicate Q3/seq: " + data.count());
    data = data.dropDuplicates("sequence").cache();
    System.out.println("- duplicate seq   : " + data.count());
    // add a property encoded feature vector
    ProteinSequenceEncoder encoder = new ProteinSequenceEncoder(data);
    data = encoder.propertyEncode();
    data.printSchema();
    data.show(25, false);
    if (args[1].equals("json")) {
        // coalesce data into a single file
        data = data.coalesce(1);
    }
    data.write().mode("overwrite").format(args[1]).save(args[0]);
    long end = System.nanoTime();
    System.out.println(TimeUnit.NANOSECONDS.toSeconds(end - start) + " sec.");
}

Example 9 with ProteinSequenceEncoder

use of edu.sdsc.mmtf.spark.ml.ProteinSequenceEncoder in project mmtf-spark by sbl-sdsc.

the class SecondaryStructureWord2VecEncoder method main.

/**
 * @param args outputFilePath outputFormat (json|parquet)
 * @throws IOException
 * @throws StructureException
 */
public static void main(String[] args) throws IOException {
    String path = MmtfReader.getMmtfReducedPath();
    if (args.length != 2) {
        System.err.println("Usage: " + SecondaryStructureWord2VecEncoder.class.getSimpleName() + " <outputFilePath> + <fileFormat>");
        System.exit(1);
    }
    long start = System.nanoTime();
    SparkConf conf = new SparkConf().setMaster("local[*]").setAppName(SecondaryStructureWord2VecEncoder.class.getSimpleName());
    JavaSparkContext sc = new JavaSparkContext(conf);
    // read MMTF Hadoop sequence file and create a non-redundant Pisces
    // subset set (<=20% seq. identity) of L-protein chains
    int sequenceIdentity = 20;
    double resolution = 3.0;
    JavaPairRDD<String, StructureDataInterface> pdb = MmtfReader.readSequenceFile(path, sc).flatMapToPair(new StructureToPolymerChains()).filter(new Pisces(sequenceIdentity, resolution));
    // get content
    int segmentLength = 11;
    Dataset<Row> data = SecondaryStructureSegmentExtractor.getDataset(pdb, segmentLength);
    // add Word2Vec encoded feature vector
    ProteinSequenceEncoder encoder = new ProteinSequenceEncoder(data);
    int n = 2;
    int windowSize = (segmentLength - 1) / 2;
    int vectorSize = 50;
    data = encoder.overlappingNgramWord2VecEncode(n, windowSize, vectorSize);
    data.printSchema();
    data.show(25, false);
    if (args[1].equals("json")) {
        // coalesce data into a single file
        data = data.coalesce(1);
    }
    data.write().mode("overwrite").format(args[1]).save(args[0]);
    long end = System.nanoTime();
    System.out.println(TimeUnit.NANOSECONDS.toSeconds(end - start) + " sec.");
}

Example 10 with ProteinSequenceEncoder

use of edu.sdsc.mmtf.spark.ml.ProteinSequenceEncoder in project mmtf-spark by sbl-sdsc.

the class SwissProtSequenceToWord2Vec2 method main.

/**
 * Generates Word2Vector models from protein sequences
 * in UniProt using overlapping n-grams.
 *
 * @param args args[0] word2VecModel output file
 * @throws IOException
 */
public static void main(String[] args) throws IOException {
    if (args.length != 1) {
        System.err.println("Usage: " + SwissProtSequenceToWord2Vec2.class.getSimpleName() + " <outputFileName>");
        System.exit(1);
    }
    long start = System.nanoTime();
    SparkSession.builder().master("local[*]").getOrCreate();
    Dataset<Row> data = UniProt.getDataset(UniProtDataset.SWISS_PROT);
    data.show(10, false);
    int segmentLength = 11;
    // add Word2Vec encoded feature vector
    ProteinSequenceEncoder encoder = new ProteinSequenceEncoder(data);
    int n = 2;
    int windowSize = (segmentLength - 1) / 2;
    int vectorSize = 50;
    data = encoder.overlappingNgramWord2VecEncode(n, windowSize, vectorSize);
    encoder.getWord2VecModel().save(args[0]);
    long end = System.nanoTime();
    System.out.println(TimeUnit.NANOSECONDS.toSeconds(end - start) + " sec.");
}