Replace BED score type with Int (#43)

* Replaced BED score type with Int
* Added a test for int score parsing
* Added CHANGES, edit the date before release

See issue JetBrains-Research/bioinf-commons#2

CHANGES

@@ -3,6 +3,16 @@ big Changelog
 
 Here you can see the full list of changes between each big release.
 
+Version 0.8.5
+-------------
+
+Released on XXXXXX
+
+Fixed
+- `ExtendedBedEntry.score` is now Int instead of Short. The reason for this is that many
+  BED file providers (e.g. MACS2, SICER) don't respect the UCSC standard which limits the score
+  to 0..1000 range, and we want to be able to parse those files.
+
 Version 0.8.4
 -------------
 

src/main/kotlin/org/jetbrains/bio/big/Bed.kt

@@ -19,7 +19,7 @@ class BedFile(val path: Path) : Iterable<BedEntry>, Closeable {
         return reader.lines().map { line ->
             val chunks = line.split('\t', limit = 4)
             BedEntry(chunks[0], chunks[1].toInt(), chunks[2].toInt(),
-                     if (chunks.size == 3) "" else chunks[3])
+                if (chunks.size == 3) "" else chunks[3])
         }.iterator()
     }
 
@@ -59,9 +59,12 @@ data class BedEntry(
      * @param delimiter Custom delimiter for malformed data
      * @param omitEmptyStrings Treat several consecutive separators as one
      */
-    fun unpack(fieldsNumber: Byte = 12, extraFieldsNumber: Int? = null,
-               delimiter: Char = '\t',
-               omitEmptyStrings: Boolean = false): ExtendedBedEntry {
+    fun unpack(
+            fieldsNumber: Byte = 12,
+            extraFieldsNumber: Int? = null,
+            delimiter: Char = '\t',
+            omitEmptyStrings: Boolean = false
+    ): ExtendedBedEntry {
 
         check(fieldsNumber in 3..12) { "Fields number expected 3..12, but was $fieldsNumber" }
 
@@ -80,7 +83,7 @@ data class BedEntry(
         val score = when {
             fieldsNumber >= 5 && it.hasNext() -> {
                 val chunk = it.next()
-                if (chunk == ".") 0 else chunk.toShort()
+                if (chunk == ".") 0 else chunk.toInt()
             }
             else -> 0
         }
@@ -147,10 +150,12 @@ data class BedEntry(
             null
         }
 
-        return ExtendedBedEntry(chrom, start, end,
-                                if (name == "") "." else name,
-                                score, strand, thickStart, thickEnd, color,
-                                blockCount, blockSizes, blockStarts, extraFields)
+        return ExtendedBedEntry(
+            chrom, start, end,
+            if (name == "") "." else name,
+            score, strand, thickStart, thickEnd, color,
+            blockCount, blockSizes, blockStarts, extraFields
+        )
     }
 
     private fun String.splitToInts(size: Int): IntArray {
@@ -179,9 +184,9 @@ data class ExtendedBedEntry(
         val end: Int,
         /** Name of feature. */
         val name: String = ".",
-        /** A number from [0, 1000] that controls shading of item. */
-        val score: Short = 0,
-
+        // UCSC defines score as an integer in range [0,1000], but almost everyone ignores the range.
+        /** Feature score */
+        val score: Int = 0,
         /** + or – or . for unknown. */
         val strand: Char = '.',
         /** The starting position at which the feature is drawn thickly. **/
@@ -237,7 +242,7 @@ data class ExtendedBedEntry(
         result = 31 * result + start
         result = 31 * result + end
         result = 31 * result + name.hashCode()
-        result = 31 * result + score
+        result = 31 * result + score.hashCode()
         result = 31 * result + strand.hashCode()
         result = 31 * result + thickStart
         result = 31 * result + thickEnd
@@ -250,18 +255,18 @@ data class ExtendedBedEntry(
     }
 
     override fun toString() = MoreObjects.toStringHelper(this)
-                .add("chrom", chrom)
-                .add("start", start).add("end", end)
-                .add("name", name)
-                .add("score", score)
-                .add("strand", strand)
-                .add("thickStart", thickStart).add("thickEnd", thickEnd)
-                .add("itemRgb", itemRgb)
-                .add("blocks", when {
-                    blockCount == 0 || blockSizes == null -> "[]"
-                    blockStarts == null -> Arrays.toString(blockSizes)
-                    else -> blockStarts.zip(blockSizes)
-                }).add("extra", extraFields?.joinToString("\t") ?: "")
+            .add("chrom", chrom)
+            .add("start", start).add("end", end)
+            .add("name", name)
+            .add("score", score)
+            .add("strand", strand)
+            .add("thickStart", thickStart).add("thickEnd", thickEnd)
+            .add("itemRgb", itemRgb)
+            .add("blocks", when {
+                blockCount == 0 || blockSizes == null -> "[]"
+                blockStarts == null -> Arrays.toString(blockSizes)
+                else -> blockStarts.zip(blockSizes)
+            }).add("extra", extraFields?.joinToString("\t") ?: "")
             .toString()
 
     /**
@@ -278,8 +283,8 @@ data class ExtendedBedEntry(
         check(fieldsNumber in 3..12) { "Fields number expected 3..12, but was $fieldsNumber" }
 
         return BedEntry(
-                chrom, start, end,
-                rest(fieldsNumber, extraFieldsNumber).joinToString(delimiter.toString())
+            chrom, start, end,
+            rest(fieldsNumber, extraFieldsNumber).joinToString(delimiter.toString())
         )
     }
 

src/test/kotlin/org/jetbrains/bio/big/BedEntryTest.kt

@@ -174,6 +174,17 @@ class BedEntryTest {
         )
     }
 
+    @Test fun unpackBed6p4IntScore() {
+        val bedEntry = BedEntry("chr1", 1, 100, ".\t40000\t+\t34.56398\t-1.00000\t4.91755\t240")
+        assertEquals(
+            ExtendedBedEntry(
+                "chr1", 1, 100, ".", 40000, '+',
+                extraFields = arrayOf("34.56398", "-1.00000", "4.91755", "240")
+            ),
+            bedEntry.unpack(fieldsNumber = 6, extraFieldsNumber = 4)
+        )
+    }
+
     @Test fun unpackBedEmptyName() {
         val bedEntry = BedEntry("chr1", 1, 100, "\t4\t+")
         assertEquals(
@@ -317,7 +328,7 @@ class BedEntryTest {
         val actualFields = (0 until 14).map { BED_ENTRY_12_P_2.getField(it, fieldsNumber, extraFieldsNumber) }
         val realExtraFieldsNumber = extraFieldsNumber ?: 2
         val expectedFields = listOf<Any?>(
-            "chr1", 10, 30, "be", 5.toShort(), '+', 15, 25, Color(15, 16, 17).rgb,
+            "chr1", 10, 30, "be", 5, '+', 15, 25, Color(15, 16, 17).rgb,
             2, intArrayOf(4, 5), intArrayOf(11, 20)
         ).slice(0 until fieldsNumber).toMutableList()
         expectedFields.addAll(listOf("val1", "4.55").slice(0 until realExtraFieldsNumber))