Add subs and grph

06_gc/gc.nim

@@ -2,7 +2,6 @@ import os
 import std/sequtils
 import std/streams
 import nre
-import std/strformat
 
 type
     DnaString = ref DnaStringObj

07_subs/sample.txt

@@ -0,0 +1,2 @@
+GATATATGCATATACTT
+ATAT

07_subs/subs.nim

@@ -0,0 +1,29 @@
+import os
+import std/streams
+import std/sequtils
+from std/strutils import join
+
+proc findNeedleInHaystack(s, t: string): seq[int] =
+    var occurrences = newSeq[int]()
+
+    for i in 0 ..< s.len:
+        var j = 0
+        while j < t.len and (i+j) < s.len and s[i+j] == t[j]:
+            inc j
+        if j == t.len:
+            occurrences.add(i+1)
+
+    return occurrences
+
+proc parseStream(stream: Stream): (string, string) =
+    let s = readLine(stream)
+    let t = readLine(stream)
+    return (s, t)
+
+let fileName = paramStr(1)
+let fileStream = newFileStream(fileName)
+if not isNil(fileStream):
+    let (s, t) = parseStream(fileStream)
+    let occurrences = findNeedleInHaystack(s, t)
+    echo occurrences.mapIt($it).join(" ")
+    fileStream.close()

08_grph/README.md

@@ -0,0 +1,3 @@
+# Overlap Graphs
+
+https://rosalind.info/problems/grph/

08_grph/grph.nim

@@ -0,0 +1,84 @@
+import os
+import std/streams
+# import re
+import nre
+import std/tables
+
+const k = 3
+
+# let fileName = paramStr(1)
+# var dna = readFile(fileName)
+
+type
+    Node* = tuple[name: string, dna: string]
+    AdjacencyList = seq[tuple[v: Node, w: Node]]
+    AffixTable* = Table[string, seq[Node]]
+
+var adjList: ref AdjacencyList = seq[tuple[v: Node, w: Node]].new()
+
+var nodesByPrefix: ref AffixTable = newTable[string, seq[Node]]()
+var nodesBySuffix: ref AffixTable = newTable[string, seq[Node]]()
+
+let firstLinePattern = re">(\w+)(?:\s+.+)?"
+
+proc parseNextNode*(stream: Stream): Node =
+    if atEnd(stream):
+        raise newException(ValueError, "Stream is at end")
+
+    let firstLine = readLine(stream)
+    let match = firstLine.match(firstLinePattern)
+    if match.isNone:
+        raise newException(ValueError, "Invalid line: " & firstLine)
+
+    let name = match.get.captures[0]
+    var dna = ""
+
+    while not atEnd(stream) and peekChar(stream) != '>':
+        dna &= readLine(stream)
+
+    (name, dna)
+
+proc addNodeToTables*(node: Node, prefixes, suffixes: ref AffixTable) =
+    let prefix = node.dna[0..k-1]
+    let suffix = node.dna[^k..^1]
+
+    if not prefixes.hasKey(prefix):
+        prefixes[prefix] = newSeq[Node]()
+    
+    prefixes[prefix].add(node)
+
+    if not suffixes.hasKey(suffix):
+        suffixes[suffix] = newSeq[Node]()
+    
+    suffixes[suffix].add(node)
+
+proc updateAdjacencyList*(node: Node, prefixes, suffixes: ref AffixTable, adjList: ref AdjacencyList) = 
+    let prefix = node.dna[0..k-1]
+    let suffix = node.dna[^k..^1]
+
+    if prefixes.hasKey(suffix):
+        for n in items(prefixes[suffix]):
+            if node.name != n.name:
+                adjList[].add((node, n))
+    
+    if suffixes.hasKey(prefix):
+        for n in items(suffixes[prefix]):
+            if node.name != n.name:
+                adjList[].add((n, node))
+
+proc processInputStream(stream: Stream) =
+    while not atEnd(stream):
+        let node = parseNextNode(stream)
+        addNodeToTables(node, nodesByPrefix, nodesBySuffix)
+        updateAdjacencyList(node, nodesByPrefix, nodesBySuffix, adjList)
+
+
+let fileName = paramStr(1)
+let fileStream = newFileStream(fileName)
+if not isNil(fileStream):
+    processInputStream(fileStream)
+
+    for t in items(adjList[]):
+        echo t.v.name & " " & $t.w.name
+    
+    fileStream.close()

08_grph/grph_test.nim

@@ -0,0 +1,46 @@
+import std/sequtils
+import std/streams
+import std/tables
+import unittest
+
+import grph
+
+suite "Overlap Graphs":
+  test "parseNextNode_emptyString_raisesException":
+    expect(ValueError):
+      discard parseNextNode(newStringStream(""))
+
+  test "parseNextNode_withInvalidLine_raisesException":
+    expect(ValueError):
+      discard parseNextNode(newStringStream("Not a valid line"))
+  
+  test "parseNextNode_withValidLine_returnsNode":
+    var stream: StringStream = newStringStream("""
+>Rosalind_0498 something
+AAATAAA
+""")
+    check parseNextNode(stream) == ("Rosalind_0498", "AAATAAA")
+  
+  test "addNodeToTables_withValidNode_addsNodeToTables":
+    var prefixes = newTable[string, seq[Node]]()
+    var suffixes = newTable[string, seq[Node]]()
+    let node = ("Rosalind_0498", "AAATGGG")
+
+    addNodeToTables(node, prefixes, suffixes)
+
+    check any(prefixes["AAA"], proc(n: Node): bool = n == node)
+    check any(suffixes["GGG"], proc(n: Node): bool = n == node)
+  
+  test "updateAdjacencyList_addsNodeCorrectly":
+    let rosalind_0498 = ("Rosalind_0498", "AAATAAA")
+    let rosalind_2391 = ("Rosalind_2391", "AAATTTT")
+    var prefixes = newTable[string, seq[Node]]()
+    var suffixes = newTable[string, seq[Node]]()
+    var adjList = seq[tuple[v: Node, w: Node]].new()
+
+    addNodeToTables(rosalind_0498, prefixes, suffixes)
+    updateAdjacencyList(rosalind_0498, prefixes, suffixes, adjList)
+    addNodeToTables(rosalind_2391, prefixes, suffixes)
+    updateAdjacencyList(rosalind_2391, prefixes, suffixes, adjList)
+    
+    check adjList[].any(proc (nodes: tuple[v: Node, w: Node]): bool = nodes.v.name == "Rosalind_0498")

08_grph/sample.txt

@@ -0,0 +1,10 @@
+>Rosalind_0498
+AAATAAA
+>Rosalind_2391
+AAATTTT
+>Rosalind_2323
+TTTTCCC
+>Rosalind_0442
+AAATCCC
+>Rosalind_5013
+GGGTGGG