README.md

ZSI-BIO-CNV

The goal of this project is to create a library with a fast and scalable implementation of the algorithms detecting CNV (Copy Number Variation) mutations in a human genome. In addition, it contains many popular features used to analyze genomic data, e.g. calculating coverage, RPKM, ZRPKM or SVD-ZRPKM values.

Technologies used:

• Apache Spark
• Hadoop-BAM

Dependencies

In order to build the project you need to have the following dependencies installed:

• JVM 1.7
• Scala 2.10.4
• SBT 0.13.5

Getting Started

1. Checkout the source using git clone https://github.com/mkopacz/zsi-bio-cnv.git
2. Build the package using sbt package from the project directory
3. Run the tests using sbt test from the project directory

If all tests are passed, the library is ready to be used. If you want to generate an API documentation run sbt doc from the project directory.

Examples

1. Calculating coverage:

val samples: Map[Int, String] = scanForSamples("path/to/bam/files")
val reads: RDD[(Int, SAMRecord)] = loadReads(sc, samples)
val bedFile: Broadcast[mutable.HashMap[Int, (Int, Int, Int)]] = sc.broadcast {
    readBedFile("path/to/bed/file")
}

val coverage: RDD[(Int, Iterable[(Int, Int)])] = {
    val counter = new CoverageCounter(sc, bedFile, reads)
    coverageToRegionCoverage(counter.calculateCoverage)
}

Check out the API documentation for more information about various options of the coverage calculation.

2. Calculating RPKM values:

// this is continuation of previous example

val rpkms: RDD[(Int, Array[Double])] = {
    val counter = new RpkmsCounter(reads, bedFile, coverage)
    counter.calculateRpkms
}

3. Calculating ZRPKM values:

// this is continuation of previous example

val zrpkms: RDD[(Int, Array[Double])] = {
    val counter = new ZrpkmsCounter(samples, rpkms, 1.0)
    counter.calculateZrpkms
}

4. Calculating SVD-ZRPKM values:

// this is continuation of previous example

val matrices: RDD[(Int, Array[Int], RealMatrix)] = {
    val counter = new SvdCounter(sc, samples, bedFile, zrpkms, 12)
    counter.calculateSvd
}

5. Making calls:

// this is continuation of previous example

val calls: RDD[(Int, Int, Int, Int, String)] = {
    val caller = new Caller(bedFile, matrices, 1.5)
    caller.call
}

6. CoNIFER algorithm:

val bedFilePath = "path/to/bed/file"
val bamFilesPath = "path/to/bam/files"

val conifer = new Conifer(sc, bedFilePath, bamFilesPath, 1.0, 1, 1.5)
conifer.calculate