Merge pull request #25 from ryan-williams/ug

upgrade tests, plugin

build.sbt

@@ -1,28 +1,26 @@
-group("org.hammerlab.genomics")
-name := "loci"
-v"2.0.4"
+subgroup("genomics", "loci")
+v"2.1.0"
 
-addSparkDeps
+spark
 
 import genomics.reference
 
 dep(
-        args4j,
-        args4s % "1.3.1",
+      case_app,
         htsjdk,
-     iterators % "2.1.0",
+     iterators % "2.2.0",
          paths % "1.5.0",
      reference % "1.4.3" + testtest,
     scalautils,
-    spark_util % "2.0.3",
+    spark_util % "2.0.4",
   string_utils % "1.2.0"
 )
 
 // Shade Guava due to use of RangeSet classes from 16.0.1 that don't exist in Spark/Hadoop's Guava 11.0.2.
 shadedDeps += guava
 
 // Rename shaded Guava classes
-shadeRenames += "com.google.common.**" → "org.hammerlab.guava.@1"
+shadeRenames += "com.google.common.**"     → "org.hammerlab.guava.@1"
 shadeRenames += "com.google.thirdparty.**" → "org.hammerlab.guava.@1"
 
 // Publish JAR that includes shaded Guava.

project/build.properties

@@ -1 +1 @@
-sbt.version=1.1.1
+sbt.version=1.1.5

project/plugins.sbt

@@ -1 +1 @@
-addSbtPlugin("org.hammerlab.sbt" % "base" % "4.4.2")
+addSbtPlugin("org.hammerlab.sbt" % "base" % "4.6.1")

src/main/scala/org/hammerlab/genomics/loci/Args.scala

@@ -0,0 +1,31 @@
+package org.hammerlab.genomics.loci
+
+import caseapp.{ HelpMessage ⇒ M }
+import org.hammerlab.genomics.loci.parsing.ParsedLoci
+import org.hammerlab.genomics.loci.parsing.ParsedLoci.loadFromPath
+import org.hammerlab.genomics.reference.ContigName.Factory
+import org.hammerlab.paths.Path
+
+/** Arguments for accepting a set of loci to restrict variant-calling to. */
+case class Args(
+  @M("If set, loci to include. Either 'all' or 'contig[:start[-end]],contig[:start[-end]],…'")
+  loci: Option[String] = None,
+
+  @M("Path to file giving loci to include")
+  lociFile: Option[Path] = None
+) {
+  def parse(
+      implicit
+      factory: Factory
+  ): Option[ParsedLoci] =
+    (loci, lociFile) match {
+      case (Some(str), Some(file)) ⇒
+        throw new IllegalArgumentException(
+          "Specify a loci string (--loci) xor file (--loci-file)"
+        )
+      case (Some(str), _) => Some(ParsedLoci(str))
+      case (_, Some(path)) => Some(loadFromPath(path))
+      case _ =>
+        None
+    }
+}

src/main/scala/org/hammerlab/genomics/loci/parsing/ParsedLoci.scala

@@ -2,7 +2,6 @@ package org.hammerlab.genomics.loci.parsing
 
 import htsjdk.variant.vcf.VCFFileReader
 import org.hammerlab.genomics.loci.VariantContext
-import org.hammerlab.genomics.loci.args.LociArgs
 import org.hammerlab.genomics.reference.ContigName.Factory
 import org.hammerlab.paths.Path
 
@@ -20,7 +19,7 @@ import scala.collection.mutable.ArrayBuffer
  * The two implementations are:
  *
  *   - [[All]]: sentinel value representing all loci on all contigs.
- *   - [[LociRanges]]: a sequence of [[LociRange]]s denoting (possibly open-ended) genomic-intervals.
+ *   - [[LociRanges]]: a sequence of [[Range]]s denoting (possibly open-ended) genomic-intervals.
  *
  * Examples:
  *
@@ -35,41 +34,21 @@ object ParsedLoci {
   def apply(lociStrs: String)(implicit factory: Factory): ParsedLoci = apply(Iterator(lociStrs))
 
   def apply(lines: Iterator[String])(implicit factory: Factory): ParsedLoci = {
-    val lociRanges = ArrayBuffer[LociRange]()
+    val lociRanges = ArrayBuffer[Range]()
     for {
       lociStrs ← lines
       lociStr ← lociStrs.replaceAll("\\s", "").split(",")
-      lociRange ← ParsedLociRange(lociStr)
+      lociRange ← ParsedRange(lociStr)
     } {
       lociRange match {
-        case AllRange ⇒ return All
-        case lociRange: LociRange ⇒
+        case ParsedRange.All ⇒ return All
+        case lociRange: Range ⇒
           lociRanges += lociRange
       }
     }
     LociRanges(lociRanges)
   }
 
-  /**
-   * Parse string representations of loci ranges, either from one string (lociOpt) or a file with one range per line
-   * (lociFileOpt), and return a [[ParsedLoci]] encapsulating the result. The latter can then be converted into a
-   * [[org.hammerlab.genomics.loci.set.LociSet]] when contig-lengths are available / have been parsed from read-sets.
-   */
-  def apply(args: LociArgs): Option[ParsedLoci] =
-    apply(
-      args.lociStrOpt,
-      args.lociFileOpt
-    )
-
-  def apply(lociStrOpt: Option[String],
-            lociFileOpt: Option[Path])(implicit factory: Factory): Option[ParsedLoci] =
-    (lociStrOpt, lociFileOpt) match {
-      case (Some(lociStr), _) => Some(ParsedLoci(lociStr))
-      case (_, Some(lociPath)) => Some(loadFromPath(lociPath))
-      case _ =>
-        None
-    }
-
   /**
    * Parse loci from the specified file.
    *
@@ -78,7 +57,7 @@ object ParsedLoci {
    *                 "chrX:5-10,chr12-10-20", etc. Whitespace is ignored.
    * @return parsed loci
    */
-  private def loadFromPath(path: Path)(implicit factory: Factory): ParsedLoci =
+  def loadFromPath(path: Path)(implicit factory: Factory): ParsedLoci =
     path.extension match {
       case "vcf" ⇒ LociRanges.fromVCF(path)
       case "loci" | "txt" ⇒ ParsedLoci(path.lines)
@@ -94,10 +73,10 @@ object ParsedLoci {
  */
 case object All extends ParsedLoci
 
-case class LociRanges(ranges: Iterable[LociRange]) extends AnyVal with ParsedLoci
+case class LociRanges(ranges: Iterable[Range]) extends AnyVal with ParsedLoci
 
 object LociRanges {
-  def apply(range: LociRange): LociRanges = apply(Iterable(range))
+  def apply(range: Range): LociRanges = apply(Iterable(range))
 
   def fromVCF(path: Path): LociRanges =
     apply(
@@ -106,7 +85,7 @@ object LociRanges {
       new VCFFileReader(path.toFile, false)
         .map {
           case VariantContext(contigName, start, end) =>
-            LociRange(contigName, start, end)
+            Range(contigName, start, end)
         }
     )
 }

src/main/scala/org/hammerlab/genomics/loci/parsing/LociRange.scala → src/main/scala/org/hammerlab/genomics/loci/parsing/Range.scala

@@ -14,9 +14,9 @@ import org.hammerlab.genomics.reference.{ ContigName, Locus }
  *   - all
  *   - none
  */
-sealed trait ParsedLociRange
+sealed trait ParsedRange
 
-object ParsedLociRange {
+object ParsedRange {
 
   val contigAndLoci = """^([\pL\pN._]+):(\pN+)(?:-(\pN*))?$""".r
   val contigOnly = """^([\pL\pN._]+)""".r
@@ -31,10 +31,10 @@ object ParsedLociRange {
    *  "chr1:10000": just chr1, position 10000; equivalent to "chr1:10000-10001".
    *  "chr1:10000-": chr1, from position 10000 to the end of chr1.
    */
-  def apply(lociRangeStr: String)(implicit factory: Factory): Option[ParsedLociRange] =
+  def apply(lociRangeStr: String)(implicit factory: Factory): Option[ParsedRange] =
     lociRangeStr.replaceAll("\\s", "") match {
       case "all" =>
-        Some(AllRange)
+        Some(All)
       case "none" | "" =>
         None
       case contigAndLoci(name, startStr, endStrOpt) =>
@@ -46,27 +46,28 @@ object ParsedLociRange {
             case None => Some(start.next)
           }
 
-        Some(LociRange(name, start, endOpt))
+        Some(Range(name, start, endOpt))
       case contigOnly(contig) =>
-        Some(LociRange(contig, Locus(0), None))
+        Some(Range(contig, Locus(0), None))
       case other =>
         throw new IllegalArgumentException(s"Couldn't parse loci range: $other")
     }
-}
 
-case object AllRange extends ParsedLociRange
+  case object All extends ParsedRange
+
+}
 
-case class LociRange(contigName: ContigName,
-                     start: Locus,
-                     endOpt: Option[Locus])
-  extends ParsedLociRange
+case class Range(contigName: ContigName,
+                 start: Locus,
+                 endOpt: Option[Locus])
+  extends ParsedRange
 
-object LociRange {
-  def apply(contigName: ContigName, start: Locus, end: Locus): LociRange =
-    LociRange(contigName, start, Some(end))
+object Range {
+  def apply(contigName: ContigName, start: Locus, end: Locus): Range =
+    Range(contigName, start, Some(end))
 
-  def apply(tuple: (ContigName, Locus, Locus)): LociRange = {
+  def apply(tuple: (ContigName, Locus, Locus)): Range = {
     val (contigName, start, end) = tuple
-    LociRange(contigName, start, Some(end))
+    Range(contigName, start, Some(end))
   }
 }

src/main/scala/org/hammerlab/genomics/loci/set/LociSet.scala

@@ -3,7 +3,7 @@ package org.hammerlab.genomics.loci.set
 import com.esotericsoftware.kryo.io.{ Input, Output }
 import com.esotericsoftware.kryo.{ Kryo, Serializer }
 import htsjdk.samtools.util.{ Interval ⇒ HTSJDKInterval }
-import org.hammerlab.genomics.loci.parsing.{ All, LociRange, LociRanges, ParsedLoci }
+import org.hammerlab.genomics.loci.parsing.{ All, Range, LociRanges, ParsedLoci }
 import org.hammerlab.genomics.reference.ContigName.Factory
 import org.hammerlab.genomics.reference.{ ContigLengths, ContigName, Interval, Locus, NumLoci, Region }
 import org.hammerlab.strings.TruncatedToString
@@ -158,7 +158,7 @@ object LociSet {
             Region(contig, Locus(0), Locus(length))
         case LociRanges(ranges) ⇒
           for {
-            LociRange(contigName, start, endOpt) ← ranges
+            Range(contigName, start, endOpt) ← ranges
             contigLengthOpt = contigLengths.get(contigName)
           } yield
             (endOpt, contigLengthOpt) match {

src/test/scala/org/hammerlab/genomics/loci/cmps.scala

@@ -0,0 +1,10 @@
+package org.hammerlab.genomics.loci
+
+import org.hammerlab.cmp.CanEq.Cmp
+import org.hammerlab.cmp.Cmp
+import org.hammerlab.genomics.reference.{ Interval, Locus }
+
+trait cmps {
+  // TODO: move this to reference repo
+  implicit def cmpInterval(implicit cmp: Cmp[(Locus, Locus)]): Cmp.Aux[Interval, cmp.Diff] = Cmp.by[(Locus, Locus), Interval](i ⇒ (i.start, i.end))(cmp)
+}

src/test/scala/org/hammerlab/genomics/loci/iterator/SkippableLociIteratorSuite.scala

@@ -24,7 +24,8 @@ class SkippableLociIteratorSuite
     )
 
   test("no skips") {
-    strs.toSeq should ===(
+    ==(
+      strs.toSeq,
       List(
         10 → "a",
         11 → "b",
@@ -49,34 +50,35 @@ class SkippableLociIteratorSuite
   test("misc skips") {
     val it = strs
     it.skipTo(15)
-    it.next() should ===(20 → "c")
+    ==(it.next(), 20 → "c")
     it.skipTo(30)
-    it.next() should ===(30 → "e")
+    ==(it.next(), 30 → "e")
     intercept[IllegalArgumentException] {
       it.skipTo(30)
     }
-    it.next() should ===(31 → "f")
+    ==(it.next(), 31 → "f")
     it.skipTo(32)
-    it.next() should ===(33 → "g")
+    ==(it.next(), 33 → "g")
     it.skipTo(34)
-    it.next() should ===(34 → "h")
+    ==(it.next(), 34 → "h")
     it.skipTo(41)
-    it.next() should ===(50 → "j")
-    it.hasNext should ===(false)
+    ==(it.next(), 50 → "j")
+    ==(it.hasNext, false)
   }
 
   test("intersect") {
-    strs.intersect(
-      new LociIterator(
-        Iterator(
-          Interval( 8, 11),
-          Interval(14, 16),
-          Interval(30, 35),
-          Interval(38, 42),
-          Interval(50, 51)
-        ).buffered
-      )
-    ).toSeq ===
+    ==(
+      strs.intersect(
+        new LociIterator(
+          Iterator(
+            Interval( 8, 11),
+            Interval(14, 16),
+            Interval(30, 35),
+            Interval(38, 42),
+            Interval(50, 51)
+          ).buffered
+        )
+      ).toSeq,
       Seq(
         10 → "a",
         30 → "e",
@@ -86,6 +88,7 @@ class SkippableLociIteratorSuite
         40 → "i",
         50 → "j"
       )
+    )
   }
 }