Benchmarking

project/Build.scala

@@ -10,7 +10,7 @@ object ApplicationBuild extends Build {
     baseDirectory in run := file("."),
     baseDirectory in Test := file("."),
 
-    scalaVersion := "2.11.7",
+    scalaVersion := "2.12.4",
     scalacOptions ++= Seq("-unchecked", "-deprecation", "-feature"),
 
     showSuccess := false,
@@ -24,8 +24,8 @@ object ApplicationBuild extends Build {
     unmanagedBase <<= baseDirectory { base => base / "lib" },
 
     libraryDependencies ++= Seq(
-      "org.scalatest" % "scalatest_2.11" % "2.2.4" % "test",
-      "com.github.scopt" %% "scopt" % "3.3.0"
+      "org.scalatest" % "scalatest_2.12" % "3.0.1" % "test",
+      "com.github.scopt" %% "scopt" % "3.7.0"
     ),
 
     mainClass in (Compile, run) := Some("tps.Main"),

scripts/compare-undirected-networks

@@ -0,0 +1,8 @@
+#!/bin/bash
+
+NETWORK_1=$1
+NETWORK_2=$2
+
+sbt \
+  "project tps-core" \
+  "run-main tps.evaluation.UndirectedNetworkComparison $NETWORK_1 $NETWORK_2"

scripts/generate-scores-and-prizes

@@ -0,0 +1,9 @@
+#!/bin/bash
+
+TIME_SERIES_FILE=$1
+MAPPING_FILE=$2
+FOLD_CHANGE_THRESHOLD=$3
+
+sbt \
+  "project tps-core" \
+  "run-main tps.evaluation.SignificanceScoreAndPrizeProducer $TIME_SERIES_FILE $MAPPING_FILE $FOLD_CHANGE_THRESHOLD"

scripts/permute-time-series-and-scores

@@ -0,0 +1,13 @@
+#!/bin/bash
+
+TIMESERIES=data/timeseries/median-time-series.tsv
+FIRSTSCORES=data/timeseries/p-values-first.tsv
+PREVSCORES=data/timeseries/p-values-prev.tsv
+
+for i in `seq 1 100`;
+do
+  OUTFOLDER="permutation-$i"
+  sbt \
+    "project tps-core" \
+    "run-main tps.evaluation.TimeSeriesPermutation $TIMESERIES $FIRSTSCORES $PREVSCORES $i $OUTFOLDER"
+done    

scripts/run-funchisq-to-sif

@@ -0,0 +1,10 @@
+#!/bin/bash
+
+NETWORK_FILE=$1
+MAPPING_FILE=$2
+P_VAL_THRESHOLD=$3
+MAX_NB_EDGES=$4
+
+sbt \
+  "project tps-core" \
+  "run-main tps.evaluation.funchisq.FunChisq2SIFConverter $NETWORK_FILE $MAPPING_FILE $P_VAL_THRESHOLD $MAX_NB_EDGES"

scripts/run-network-analysis

@@ -0,0 +1,9 @@
+#!/bin/bash
+
+for reference in data/resources/reference/*.tsv
+do
+  echo "Evaluating against $reference"
+  sbt \
+    "project tps-core" \
+    "run-main tps.evaluation.NetworkReferenceComparison $reference $*"
+done

scripts/run-one-to-many-network-comparison

@@ -0,0 +1,5 @@
+#!/bin/bash
+
+sbt \
+  "project tps-core" \
+  "run-main tps.evaluation.DirectedNetworkComparison $*"

scripts/run-overlap-analysis

@@ -0,0 +1,13 @@
+#!/bin/bash
+
+timeseries=data/timeseries/median-time-series.tsv
+firstscores=data/timeseries/p-values-first.tsv
+prevscores=data/timeseries/p-values-prev.tsv
+mapping=data/timeseries/peptide-mapping.tsv
+
+for reference in data/resources/reference/*.tsv
+do
+  sbt \
+    "project tps-core" \
+    "run-main tps.evaluation.OverlapAnalysis $reference $timeseries $mapping $firstscores $prevscores"
+done

scripts/run-permuted-time-series

@@ -0,0 +1,15 @@
+#!/bin/bash
+
+for i in `seq 1 100`;
+do
+  ./scripts/run \
+    --network data/networks/input-network.tsv \
+    --timeseries data/timeseries/permuted-time-series/permutation-$i/permuted-time-series.tsv \
+    --firstscores data/timeseries/permuted-time-series/permutation-$i/permuted-first-scores.tsv \
+    --prevscores data/timeseries/permuted-time-series/permutation-$i/permuted-prev-scores.tsv \
+    --partialmodel data/resources/kinase-substrate-interactions.sif \
+    --peptidemap data/timeseries/peptide-mapping.tsv \
+    --source EGF_HUMAN \
+    --threshold 0.01 \
+    --outfolder permuted-time-series-run-$i
+done

scripts/run-txn

@@ -0,0 +1,68 @@
+#!/bin/bash
+
+NETWORKFOLDER=data/networks/evaluation
+KINSUB=data/resources/kinase-substrate-interactions.sif
+
+TXN_ALL_UNDIRECTED=$NETWORKFOLDER/TXN-ALL-undirected.tsv
+TXN_ALL_DIRECTED=$NETWORKFOLDER/TXN-ALL-directed.sif
+TXN_1F_UNDIRECTED=$NETWORKFOLDER/TXN-1F-undirected.tsv
+TXN_1F_DIRECTED=$NETWORKFOLDER/TXN-1F-directed.sif
+
+function run_with_base_args() {
+  ./scripts/run \
+    --timeseries data/timeseries/median-time-series.tsv \
+    --firstscores data/timeseries/p-values-first.tsv \
+    --prevscores data/timeseries/p-values-prev.tsv \
+    --peptidemap data/timeseries/peptide-mapping.tsv \
+    --source EGFR_HUMAN \
+    --threshold 0.01 \
+    "$@"
+}
+
+run_with_base_args \
+  --outlabel TXN-ALL-directed-topo-kinsub \
+  --network $TXN_ALL_UNDIRECTED \
+  --partialmodel $KINSUB \
+  --partialmodel $TXN_ALL_DIRECTED \
+  --no-temporality
+
+run_with_base_args \
+  --outlabel TXN-ALL-directed-topo-kinsub-temp \
+  --network $TXN_ALL_UNDIRECTED \
+  --partialmodel $KINSUB \
+  --partialmodel $TXN_ALL_DIRECTED
+
+run_with_base_args \
+  --outlabel TXN-1F-directed-topo-kinsub \
+  --network $TXN_1F_UNDIRECTED \
+  --partialmodel $KINSUB \
+  --partialmodel $TXN_ALL_DIRECTED \
+  --no-temporality
+
+run_with_base_args \
+  --outlabel TXN-1F-directed-topo-kinsub-temp \
+  --network $TXN_1F_UNDIRECTED \
+  --partialmodel $KINSUB \
+  --partialmodel $TXN_ALL_DIRECTED
+
+run_with_base_args \
+  --outlabel TXN-ALL-undirected-topo-kinsub \
+  --network $TXN_ALL_UNDIRECTED \
+  --partialmodel $KINSUB \
+  --no-temporality
+
+run_with_base_args \
+  --outlabel TXN-ALL-undirected-topo-kinsub-temp \
+  --network $TXN_ALL_UNDIRECTED \
+  --partialmodel $KINSUB
+
+run_with_base_args \
+  --outlabel TXN-1F-undirected-topo-kinsub \
+  --network $TXN_1F_UNDIRECTED \
+  --partialmodel $KINSUB \
+  --no-temporality
+
+run_with_base_args \
+  --outlabel TXN-1F-undirected-topo-kinsub-temp \
+  --network $TXN_1F_UNDIRECTED \
+  --partialmodel $KINSUB

scripts/run-txn-graph-creation

@@ -0,0 +1,6 @@
+#!/bin/bash
+
+TXNFILE="data/networks/comparison/TXN-ALL.edges"
+sbt \
+  "project tps-core" \
+  "run-main tps.evaluation.txn.GraphCreation $TXNFILE 1.0"

tps-core/src/main/scala/tps/Graphs.scala

@@ -1,7 +1,5 @@
 package tps
 
-import tps.util.LogUtils._
-
 object Graphs {
 
   case class Vertex(id: String) extends Serializable {

tps-core/src/main/scala/tps/PeptideProteinMappingParser.scala

@@ -1,7 +1,5 @@
 package tps
 
-import tps.util.FileUtils._
-
 object PeptideProteinMappingParser {
   def run(f: java.io.File): Map[String, Set[String]] = {
     val data = new TSVSource(f).data

tps-core/src/main/scala/tps/ReferenceParser.scala

@@ -7,10 +7,11 @@ object ReferenceParser {
   def run(f: java.io.File): (SignedDirectedGraph, Map[Edge, String]) = {
     val data = new TSVSource(f, noHeaders = false).data
 
-
     var evidencePerEdge: Map[Edge, String] = Map.empty
 
-    val tuples = data.tuples.collect{ case tuple if tuple.size >= 7 =>
+    val tuples = data.tuples.map{ tuple =>
+      assert(tuple.size >= 6)
+
       val Seq(src, tgt, lra, lri, rla, rli, rest @ _*) = tuple
       val edge = lexicographicEdge(src, tgt)
 
@@ -21,6 +22,7 @@ object ReferenceParser {
       def labelValue(l: String): Boolean = l match {
         case "true" => true
         case "false" => false
+        case _ => sys.error(s"error parsing $l in $tuple")
       }
 
       if (labelValue(lra)) originalLabels += ActiveEdge(Forward, Activating)

tps-core/src/main/scala/tps/SignedDirectedGraphOps.scala

@@ -124,10 +124,64 @@ object SignedDirectedGraphOps {
     relevant.size
   }
 
+  def activeDirections(
+    labels: Set[SignedDirectedEdgeLabel]
+  ): Set[EdgeDirection] = {
+    labels collect {
+      case ActiveEdge(dir, sign) => dir
+    }
+  }
+
+  def activeEdges(labels: Set[SignedDirectedEdgeLabel]): Set[ActiveEdge] = {
+    labels collect {
+      case ae: ActiveEdge => ae
+    }
+  }
+
+  def sameUniqueDirection(
+    ls1: Set[SignedDirectedEdgeLabel], ls2: Set[SignedDirectedEdgeLabel]
+  ): Boolean = {
+    val activeDir1 = activeDirections(ls1)
+    val activeDir2 = activeDirections(ls2)
+
+    activeDir1.size == 1 && activeDir2.size == 1 && activeDir1 == activeDir2
+  }
+
+  def sameUniqueActiveEdge(
+    ls1: Set[SignedDirectedEdgeLabel], ls2: Set[SignedDirectedEdgeLabel]
+  ): Boolean = {
+    val activeE1 = activeEdges(ls1)
+    val activeE2 = activeEdges(ls2)
+
+    activeE1.size == 1 && activeE2.size == 1 && activeE1 == activeE2
+  }
+
+  def confictingUniqueActiveEdge(
+    ls1: Set[SignedDirectedEdgeLabel], ls2: Set[SignedDirectedEdgeLabel]
+  ): Boolean = {
+    val activeE1 = activeEdges(ls1)
+    val activeE2 = activeEdges(ls2)
+
+    activeE1.size == 1 && activeE2.size == 1 && activeE1 != activeE2
+  }
+
+  def conflictingUniqueActiveDirection(
+    ls1: Set[SignedDirectedEdgeLabel], ls2: Set[SignedDirectedEdgeLabel]
+  ): Boolean = {
+    val activeDir1 = activeDirections(ls1)
+    val activeDir2 = activeDirections(ls2)
+
+    activeDir1.size == 1 && activeDir2.size == 1 && activeDir1 != activeDir2
+  }
+
   def oneActiveDirection(es: Set[SignedDirectedEdgeLabel]): Boolean = {
     onlyForward(es) || onlyBackward(es)
   }
 
+  def oneActiveEdge(es: Set[SignedDirectedEdgeLabel]): Boolean = {
+    activeEdges(es).size == 1
+  }
+
   def onlyBackward(es: Set[SignedDirectedEdgeLabel]): Boolean = {
     canBeBackward(es) && !canBeForward(es)
   }

tps-core/src/main/scala/tps/TimeSeries.scala

@@ -1,9 +1,5 @@
 package tps
 
-import Graphs._
-
-import tps.util.StringUtils
-
 case class Profile(id: String, values: Seq[Option[Double]])
 
 case class TimeSeries(