rdd
The ADAMContext provides functions on top of a SparkContext for loading genomic data.
Argument configuration for saving any output format.
An abstract class that extends GenomicRDD and where the underlying data are Avro IndexedRecords.
An abstract class describing a GenomicRDD where:
An abstract class describing a GenomicRDD where:
* The data are Avro IndexedRecords. * The data are associated to record groups (i.e., they are reads or fragments).
Partition a genome into a set of bins.
Partition a genome into a set of bins.
Note that this class will not tolerate invalid input, so filter in advance if you use it.
The size of each bin in nucleotides
A map containing the length of each contig
A trait describing a GenomicRDD that also supports the Spark SQL APIs.
GenomicPositionPartitioner partitions ReferencePosition objects into separate, spatially-coherent regions of the genome.
GenomicPositionPartitioner partitions ReferencePosition objects into separate, spatially-coherent regions of the genome.
This can be used to organize genomic data for computation that is spatially distributed (e.g. GATK and Queue's "scatter-and-gather" for locus-parallelizable walkers).
The number of equally-sized regions into which the total genomic space is partitioned; the total number of partitions is numParts + 1, with the "+1" resulting from one extra partition that is used to capture null or UNMAPPED values of the ReferencePosition type.
a map relating sequence-name to length and indicating the set and length of all extant sequences in the genome.
A trait that wraps an RDD of genomic data with helpful metadata.
A trait that wraps an RDD of genomic data with helpful metadata.
The type of the data in the wrapped RDD.
The type of this GenomicRDD.
A partitioner for ReferenceRegion-keyed data.
A partitioner for ReferenceRegion-keyed data.
The number of bases per partition.
A map between contig names and contig lengths.
If true, use the start position (instead of the end position) to decide which partition a key belongs to.
Formats data going into a pipe to an invoked process.
Formats data going into a pipe to an invoked process.
The type of records being formatted.
A trait for singleton objects that build an InFormatter from a GenomicRDD.
A trait for singleton objects that build an InFormatter from a GenomicRDD.
Often, when creating an outputstream, we need to add metadata to the output that is not attached to individual records. An example of this is writing a header with contig/read group/format info, as is done with SAM/BAM/VCF.
The type of the records this InFormatter writes out.
The type of the GenomicRDD this companion object understands.
The type of InFormatter this companion object creates.
Implements an inner region join where the left side of the join is broadcast.
Performs an inner region join, followed logically by grouping by the right value.
Performs an inner region join, followed logically by grouping by the right value. This is implemented without any shuffling; the join naturally returns values on the left grouped by the right value.
An abstract class that extends the MultisampleGenomicRDD trait, where the data are Avro IndexedRecords.
A trait describing a GenomicRDD with data from multiple samples.
Deserializes data coming out of a pipe from an invoked process.
Deserializes data coming out of a pipe from an invoked process.
The type of records being formatted.
Repartitions objects that are keyed by a ReferencePosition or ReferenceRegion into a single partition per contig.
A trait describing a join in the genomic coordinate space between two RDDs where the values are keyed by a ReferenceRegion.
A trait describing a join in the genomic coordinate space between two RDDs where the values are keyed by a ReferenceRegion.
The type of the left RDD.
The type of the right RDD.
The type of data yielded by the left RDD at the output of the join. This may not match T if the join is an outer join, etc.
The type of data yielded by the right RDD at the output of the join.
Implements a right outer region join where the left side of the join is broadcast.
Performs a right outer region join, followed logically by grouping by the right value.
Performs a right outer region join, followed logically by grouping by the right value. This is implemented without any shuffling; the join naturally returns values on the left grouped by the right value. In this implementation, empty collections on the left side of the join are kept.
A trait describing join implementations that are based on a sort-merge join.
A trait describing join implementations that are based on a sort-merge join.
The type of the left RDD.
The type of the right RDD.
The type of data yielded by the left RDD at the output of the join. This may not match T if the join is an outer join, etc.
The type of data yielded by the right RDD at the output of the join.
Implements a shuffle free (broadcast) region join.
Implements a shuffle free (broadcast) region join.
The broadcast values are stored in a sorted array. It was going to be an ensemble of interval trees, but, that didn't work out.
This singleton provides an implicit conversion from a SparkContext to the ADAMContext, as well as implicit functions for the Pipe API.
Helper for creating genomic position partitioners.
Helper object for creating GenomicRegionPartitioners.
An abstract class that extends GenomicRDD and where the underlying data are Avro IndexedRecords. This abstract class provides methods for saving to Parquet, and provides hooks for writing the metadata.