Genomic Data Loading

import org.bdgenomics.adam.api.java.JavaADAMContext;
import org.bdgenomics.adam.rdd.ADAMContext;
import htsjdk.samtools.ValidationStringency;

/**
 * Java-friendly wrapper for ADAM Context providing genomic data loading capabilities
 */
class JavaADAMContext {
    /**
     * Creates a JavaADAMContext wrapping the provided ADAMContext
     * @param ac The ADAMContext to wrap
     */
    JavaADAMContext(ADAMContext ac);
    
    /**
     * Returns the Java Spark Context associated with this context
     * @return JavaSparkContext for Spark operations
     */
    JavaSparkContext getSparkContext();
}

/**
 * Load alignment records with automatic format detection
 * Supports: .bam/.cram/.sam (BAM/CRAM/SAM), .fa/.fasta (FASTA), 
 * .fq/.fastq (FASTQ), .ifq (interleaved FASTQ)
 * Falls back to Parquet + Avro for unrecognized extensions
 * @param pathName Path to alignment file(s). Supports globs and directories
 * @return AlignmentRecordRDD containing reads, sequence dictionary, and record groups
 */
AlignmentRecordRDD loadAlignments(String pathName);

/**
 * Load alignment records with validation stringency control
 * @param pathName Path to alignment file(s)
 * @param stringency Validation strictness (LENIENT, SILENT, STRICT)
 * @return AlignmentRecordRDD containing reads, sequence dictionary, and record groups
 */
AlignmentRecordRDD loadAlignments(String pathName, ValidationStringency stringency);

/**
 * Load nucleotide contig fragments from reference sequences
 * Supports: .fa/.fasta (FASTA format)
 * Falls back to Parquet + Avro for other extensions
 * @param pathName Path to reference file(s). Supports globs and directories for FASTA
 * @return NucleotideContigFragmentRDD containing reference sequences
 */
NucleotideContigFragmentRDD loadContigFragments(String pathName);

/**
 * Load reference sequences into broadcastable format
 * Supports: .2bit files directly, other formats via loadContigFragments
 * @param pathName Path to reference file (no globs/directories for 2bit)
 * @return ReferenceFile for broadcast operations
 */
ReferenceFile loadReferenceFile(String pathName);

/**
 * Load reference sequences with custom maximum fragment length
 * @param pathName Path to reference file
 * @param maximumLength Maximum fragment length (default 10000L, avoid >1e9)
 * @return ReferenceFile for broadcast operations
 */
ReferenceFile loadReferenceFile(String pathName, Long maximumLength);

/**
 * Load fragments from alignment data
 * Supports: .bam/.cram/.sam (BAM/CRAM/SAM), .ifq (interleaved FASTQ)
 * Falls back to Parquet + Avro for other extensions
 * @param pathName Path to fragment file(s). Supports globs and directories
 * @return FragmentRDD containing paired-end fragments
 */
FragmentRDD loadFragments(String pathName);

/**
 * Load fragments with validation stringency control
 * @param pathName Path to fragment file(s)
 * @param stringency Validation strictness for BAM/CRAM/SAM and FASTQ formats
 * @return FragmentRDD containing paired-end fragments
 */
FragmentRDD loadFragments(String pathName, ValidationStringency stringency);

/**
 * Load genomic features from annotation files
 * Supports: .bed (BED6/12), .gff3 (GFF3), .gtf/.gff (GTF/GFF2), 
 * .narrow[pP]eak (NarrowPeak), .interval_list (IntervalList)
 * Falls back to Parquet + Avro for other extensions
 * @param pathName Path to feature file(s). Supports globs and directories
 * @return FeatureRDD containing genomic annotations
 */
FeatureRDD loadFeatures(String pathName);

/**
 * Load features with validation stringency control
 * @param pathName Path to feature file(s)
 * @param stringency Validation strictness for supported text formats
 * @return FeatureRDD containing genomic annotations
 */
FeatureRDD loadFeatures(String pathName, ValidationStringency stringency);

/**
 * Load features and convert to coverage data
 * Coverage values are stored in the score field of Feature records
 * Supports same formats as loadFeatures
 * @param pathName Path to coverage file(s). Supports globs and directories
 * @return CoverageRDD containing coverage depth information
 */
CoverageRDD loadCoverage(String pathName);

/**
 * Load coverage data with validation stringency control
 * @param pathName Path to coverage file(s)
 * @param stringency Validation strictness for supported text formats
 * @return CoverageRDD containing coverage depth information
 */
CoverageRDD loadCoverage(String pathName, ValidationStringency stringency);

/**
 * Load genotype calls from variant files
 * Supports: .vcf/.vcf.gz/.vcf.bgzf/.vcf.bgz (VCF format)
 * Falls back to Parquet + Avro for other extensions
 * @param pathName Path to genotype file(s). Supports globs and directories for VCF
 * @return GenotypeRDD containing sample genotype calls
 */
GenotypeRDD loadGenotypes(String pathName);

/**
 * Load genotypes with validation stringency control
 * @param pathName Path to genotype file(s)
 * @param stringency Validation strictness for VCF format
 * @return GenotypeRDD containing sample genotype calls
 */
GenotypeRDD loadGenotypes(String pathName, ValidationStringency stringency);

/**
 * Load variant records from variant files
 * Supports: .vcf/.vcf.gz/.vcf.bgzf/.vcf.bgz (VCF format)
 * Falls back to Parquet + Avro for other extensions
 * @param pathName Path to variant file(s). Supports globs and directories for VCF
 * @return VariantRDD containing genetic variations
 */
VariantRDD loadVariants(String pathName);

/**
 * Load variants with validation stringency control
 * @param pathName Path to variant file(s)
 * @param stringency Validation strictness for VCF format
 * @return VariantRDD containing genetic variations
 */
VariantRDD loadVariants(String pathName, ValidationStringency stringency);

JavaADAMContext jac = new JavaADAMContext(adamContext);

// Load BAM file with default settings
AlignmentRecordRDD alignments = jac.loadAlignments("sample.bam");

// Load with strict validation
AlignmentRecordRDD strictAlignments = jac.loadAlignments("sample.bam", 
    ValidationStringency.STRICT);

// Load compressed FASTQ
AlignmentRecordRDD fastqReads = jac.loadAlignments("reads.fastq.gz");

// Load reference genome
NucleotideContigFragmentRDD reference = jac.loadContigFragments("hg38.fa");

// Create broadcastable reference
ReferenceFile refFile = jac.loadReferenceFile("hg38.2bit");

// Load genomic annotations
FeatureRDD genes = jac.loadFeatures("gencode.gtf");
CoverageRDD coverage = jac.loadCoverage("sample.bed");

// Load VCF file
VariantRDD variants = jac.loadVariants("variants.vcf.gz");
GenotypeRDD genotypes = jac.loadGenotypes("variants.vcf.gz");

// With lenient validation for problematic files
VariantRDD lenientVariants = jac.loadVariants("problematic.vcf", 
    ValidationStringency.LENIENT);

/**
 * Validation stringency enumeration for controlling file format validation
 */
enum ValidationStringency {
    STRICT,   // Fail on any format violations
    LENIENT,  // Warn on format issues but continue processing
    SILENT    // Ignore format violations silently
}