diff --git a/README.md b/README.md new file mode 100644 --- /dev/null +++ b/README.md @@ -0,0 +1,8 @@ +ViennaRNAParser [![Hackage](https://img.shields.io/hackage/v/ViennaRNAParser.svg)](https://hackage.haskell.org/package/ViennaRNAParser) [![Build Status](https://travis-ci.org/eggzilla/ViennaRNAParser.svg)](https://travis-ci.org/eggzilla/ViennaRNAParser) +============= + +Currently contains parsers and datatypes for: RNAalifold, RNAdistance, RNAcofold, RNAfold, RNAplex, RNAup, RNAz. + +For more information on the ViennaRNA package refer to . + +The libary is tested with Version 2.1.9 of the ViennaRNA package. diff --git a/ViennaRNAParser.cabal b/ViennaRNAParser.cabal --- a/ViennaRNAParser.cabal +++ b/ViennaRNAParser.cabal @@ -5,10 +5,13 @@ -- PVP summary: +-+------- breaking API changes -- | | +----- non-breaking API additions -- | | | +--- code changes with no API change -version: 1.0.1 +version: 1.2.0 synopsis: Libary for parsing ViennaRNA package output -description: Currently contains parsers and datatypes for: RNAplex, RNAz. The libary is tested with Version >2.0 of the ViennaRNA package. - +description: Currently contains parsers and datatypes for: RNAalifold, RNAdistance, RNAcofold, RNAfold, RNAplex, RNAup, RNAz. + . + For more information on the ViennaRNA package refer to . + . + The libary is tested with Version 2.1.9 of the ViennaRNA package. license: GPL license-file: LICENSE author: Florian Eggenhofer @@ -18,19 +21,24 @@ build-type: Simple cabal-version: >=1.8 +extra-source-files: + README.md changelog + source-repository head type: git location: https://github.com/eggzilla/ViennaRNAParser source-repository this type: git - location: https://github.com/eggzilla/ViennaRNAParser/tree/1.0.1 - tag: 1.0.1 + location: https://github.com/eggzilla/ViennaRNAParser/tree/1.2.0 + tag: 1.2.0 library -- Modules exported by the library. exposed-modules: Bio.ViennaRNAParser - other-modules: Bio.RNAfoldData + Bio.RNAcofoldData + Bio.RNAcofoldParser + Bio.RNAfoldData Bio.RNAfoldParser Bio.RNAplexData Bio.RNAplexParser @@ -47,9 +55,13 @@ ghc-options: -Wall -O2 -fno-warn-unused-do-bind -- Other library packages from which modules are imported. - build-depends: base >=4.5 && <5, parsec + build-depends: base >=4.5 && <5, parsec>=3.1.9, process -- Directories containing source files. hs-source-dirs: src - - + +test-suite hspec + build-depends: base, parsec, ViennaRNAParser, hspec>=1.8 + hs-source-dirs: test-suite + main-is: Spec.hs + type: exitcode-stdio-1.0 diff --git a/changelog b/changelog new file mode 100644 --- /dev/null +++ b/changelog @@ -0,0 +1,11 @@ +-*-change-log-*- +1.2.0 Florian Eggenhofer 1. July 2015 + * Added Hspec test-suite + * Added RNAcofold parser +1.1.0 Florian Eggenhofer 2. June 2015 + * Exposed submodules for imports, added systemcalls for RNAalifold, + RNAfold, RNAz +1.0.1 Florian Eggenhofer 1. March 2015 + * Added Travis CI support + * Added -Wall and -O2 compiler options + * Added RNAfold, RNAalifold, RNAdistance, RNAup parsers diff --git a/src/Bio/RNAalifoldData.hs b/src/Bio/RNAalifoldData.hs --- a/src/Bio/RNAalifoldData.hs +++ b/src/Bio/RNAalifoldData.hs @@ -4,7 +4,7 @@ module Bio.RNAalifoldData where -- | Data structure corrsponds with alifoldoutput redirected to file. Output written directly to commandline also contains number of sequences and length of consensus sequence -data RNAalifoldOutput = RNAalifoldOutput +data RNAalifold = RNAalifold { alignmentConsensusSequence :: String, alignmentConsensusDotBracket :: String, diff --git a/src/Bio/RNAalifoldParser.hs b/src/Bio/RNAalifoldParser.hs --- a/src/Bio/RNAalifoldParser.hs +++ b/src/Bio/RNAalifoldParser.hs @@ -2,6 +2,7 @@ -- For more information on RNAalifold consult: <> module Bio.RNAalifoldParser ( + systemRNAalifold, parseRNAalifold, readRNAalifold, module Bio.RNAalifoldData @@ -9,18 +10,26 @@ import Bio.RNAalifoldData import Text.ParserCombinators.Parsec +import System.Process +import System.Exit +import qualified Control.Exception.Base as CE readDouble :: String -> Double readDouble = read +--- | Run external RNAalifold command and read the output into the corresponding datatype +systemRNAalifold :: String -> String -> String -> IO ExitCode +systemRNAalifold options inputFilePath outputFilePath = system ("RNAalifold " ++ options ++ " < " ++ inputFilePath ++ " > " ++ outputFilePath) + -- | Parse the consenus of RNAz results -genParserRNAalifold :: GenParser Char st RNAalifoldOutput +genParserRNAalifold :: GenParser Char st RNAalifold genParserRNAalifold = do _sequence <- many1 (noneOf "\n") newline secondaryStructure <- many1 (oneOf "&().,") string (" (") - foldingEnergy <- many1 (noneOf " ") + optional space + foldingEnergy <- many1 (oneOf "-.1234567890") string " =" many1 space initialFoldingEnergy <- many1 (noneOf " ") @@ -30,13 +39,15 @@ string (")") many1 space eof - return $ RNAalifoldOutput _sequence secondaryStructure (readDouble foldingEnergy) (readDouble initialFoldingEnergy) (readDouble covarianceContributionEnergy) + return $ RNAalifold _sequence secondaryStructure (readDouble foldingEnergy) (readDouble initialFoldingEnergy) (readDouble covarianceContributionEnergy) -- | parse RNAalifold output from input string -parseRNAalifold :: [Char] -> Either ParseError RNAalifoldOutput +parseRNAalifold :: [Char] -> Either ParseError RNAalifold parseRNAalifold input = parse genParserRNAalifold "genParseRNAalifold" input -- | parse RNAalifold output from input filePath -readRNAalifold :: String -> IO (Either ParseError RNAalifoldOutput) -readRNAalifold filePath = parseFromFile genParserRNAalifold filePath +readRNAalifold :: String -> IO (Either ParseError RNAalifold) +readRNAalifold filePath = do + parsedFile <- parseFromFile genParserRNAalifold filePath + CE.evaluate parsedFile diff --git a/src/Bio/RNAcofoldData.hs b/src/Bio/RNAcofoldData.hs new file mode 100644 --- /dev/null +++ b/src/Bio/RNAcofoldData.hs @@ -0,0 +1,14 @@ +-- | This module contains a data structure for RNAcofold output +-- For more information on RNAcofold consult: + +module Bio.RNAcofoldData where +-- | Data structure for cofolding of 2 nucleic acid sequences +data RNAcofold = RNAcofold + { + coFoldSequence1 :: String, + coFoldSequence2 :: String, + coFoldSecondaryStructure1 :: String, + coFoldSecondaryStructure2 :: String, + coFoldingEnergy :: Double + } + deriving (Show, Eq) diff --git a/src/Bio/RNAcofoldParser.hs b/src/Bio/RNAcofoldParser.hs new file mode 100644 --- /dev/null +++ b/src/Bio/RNAcofoldParser.hs @@ -0,0 +1,49 @@ +-- | Parse RNAfold output +-- For more information on RNAcofold consult: + +module Bio.RNAcofoldParser ( + systemRNAcofold, + parseRNAcofold, + readRNAcofold, + module Bio.RNAcofoldData + ) where + +import Bio.RNAcofoldData +import Text.ParserCombinators.Parsec +import System.Process +import System.Exit +import qualified Control.Exception.Base as CE + +-- | Run external RNAcofold command +systemRNAcofold :: String -> String -> IO ExitCode +systemRNAcofold inputFilePath outputFilePath = system ("RNAcofold <" ++ inputFilePath ++ " >" ++ outputFilePath) + +readDouble :: String -> Double +readDouble = read + +-- | Parse the consenus of RNAcofold results +genParserRNAcofold :: GenParser Char st RNAcofold +genParserRNAcofold = do + _coFoldSequence1 <- many1 (oneOf "NATUGCatugc") + string "&" + _coFoldSequence2 <- many1 (oneOf "NATUGCatugc") + newline + _coFoldSecondaryStructure1 <- many1 (oneOf "().,") + string "&" + _coFoldSecondaryStructure2 <- many1 (oneOf "().,") + space + string ("(") + optional space + _coFoldingEnergy <- many1 (oneOf "-1234567890.") + string (")") + return $ RNAcofold _coFoldSequence1 _coFoldSequence2 _coFoldSecondaryStructure1 _coFoldSecondaryStructure2 (readDouble _coFoldingEnergy) + +-- | parse RNAcofold output from input string +parseRNAcofold :: [Char] -> Either ParseError RNAcofold +parseRNAcofold input = parse genParserRNAcofold "genParseRNAcofold" input + +-- | parse RNAcofold output from input filePath +readRNAcofold :: String -> IO (Either ParseError RNAcofold) +readRNAcofold filePath = do + parsedFile <- parseFromFile genParserRNAcofold filePath + CE.evaluate parsedFile diff --git a/src/Bio/RNAfoldData.hs b/src/Bio/RNAfoldData.hs --- a/src/Bio/RNAfoldData.hs +++ b/src/Bio/RNAfoldData.hs @@ -1,9 +1,8 @@ --- | This module contains a hierarchical data structure for --- RNAplex output --- For more information on RNAplex consult: +-- | This module contains a data structure for RNAfold output +-- For more information on RNAplex consult: module Bio.RNAfoldData where --- | Data structure for individual interaction between a target and query nucleic acid seqence +-- | Data structure for RNAfold data RNAfold = RNAfold { sequenceIdentifier :: String, diff --git a/src/Bio/RNAfoldParser.hs b/src/Bio/RNAfoldParser.hs --- a/src/Bio/RNAfoldParser.hs +++ b/src/Bio/RNAfoldParser.hs @@ -1,7 +1,8 @@ -- | Parse RNAfold output --- For more information on RNAplex consult: <> +-- For more information on RNAfold consult: module Bio.RNAfoldParser ( + systemRNAfold, parseRNAfold, readRNAfold, module Bio.RNAfoldData @@ -9,7 +10,14 @@ import Bio.RNAfoldData import Text.ParserCombinators.Parsec +import System.Process +import System.Exit +import qualified Control.Exception.Base as CE +-- | Run external RNAfold command and read the output into the corresponding datatype +systemRNAfold :: String -> String -> IO ExitCode +systemRNAfold inputFilePath outputFilePath = system ("RNAfold --noPS <" ++ inputFilePath ++ " >" ++ outputFilePath) + readDouble :: String -> Double readDouble = read @@ -34,4 +42,6 @@ -- | parse RNAfold output from input filePath readRNAfold :: String -> IO (Either ParseError RNAfold) -readRNAfold filePath = parseFromFile genParserRNAfold filePath +readRNAfold filePath = do + parsedFile <- parseFromFile genParserRNAfold filePath + CE.evaluate parsedFile diff --git a/src/Bio/RNAplexParser.hs b/src/Bio/RNAplexParser.hs --- a/src/Bio/RNAplexParser.hs +++ b/src/Bio/RNAplexParser.hs @@ -46,6 +46,7 @@ _queryDuplexEnd <- many1 digit many1 space char '(' + optional space _duplexEnergy <- many1 (noneOf (" )")) optional space optional (char '=') diff --git a/src/Bio/RNAupData.hs b/src/Bio/RNAupData.hs --- a/src/Bio/RNAupData.hs +++ b/src/Bio/RNAupData.hs @@ -28,6 +28,6 @@ upTargetAccessibility :: Maybe Double, upQuerySequence :: String, upTargetSequence :: String, - upOutputFileName :: String + upOutputFileName :: Maybe String } deriving (Show, Eq) diff --git a/src/Bio/RNAupParser.hs b/src/Bio/RNAupParser.hs --- a/src/Bio/RNAupParser.hs +++ b/src/Bio/RNAupParser.hs @@ -71,9 +71,16 @@ char '&' targetSequence <- many1 (noneOf ("\n")) newline - _upOutputFileName <- many1 (noneOf ("\n")) - newline + _upOutputFileName <- optionMaybe (try parseRNAupFileName) return $ RNAupInteractionRegion upsecondaryStructure (readInt uptargetDuplexBegin) (readInt uptargetDuplexEnd) (readInt upqueryDuplexBegin) (readInt upqueryDuplexEnd) (readDouble upduplexEnergy) (liftM readDouble upduplexEnergyWithoutAccessiblity) (liftM readDouble upqueryAccessiblity) (liftM readDouble uptargetAccessibility) querySequence targetSequence _upOutputFileName + +-- | Parse a RNAupFileName +parseRNAupFileName :: GenParser Char st String +parseRNAupFileName = do + string "RNAup output in file: " + fileName <- many1 (noneOf ("\n>")) + newline + return fileName -- | parse RNAupOutput from input string parseRNAup :: [Char] -> Either ParseError [RNAupInteraction] diff --git a/src/Bio/RNAzData.hs b/src/Bio/RNAzData.hs --- a/src/Bio/RNAzData.hs +++ b/src/Bio/RNAzData.hs @@ -6,7 +6,7 @@ --import Biobase.RNA -- | All elements of RNAz output are contained in this datatype -data RNAzOutput = RNAzOutput +data RNAz = RNAz { version :: String, sequenceNumber :: Int, columnNumber :: Int, diff --git a/src/Bio/RNAzParser.hs b/src/Bio/RNAzParser.hs --- a/src/Bio/RNAzParser.hs +++ b/src/Bio/RNAzParser.hs @@ -3,6 +3,7 @@ -- | Parse RNAz output -- For more information on RNAz consult: Double readDouble = read @@ -17,9 +21,13 @@ readInt :: String -> Int readInt = read --- | Parse the input as RNAzOutput datatype -parseRNAzOutput :: GenParser Char st RNAzOutput -parseRNAzOutput = do +-- | Run external RNAz command and read the output into the corresponding datatype +systemRNAz :: String -> String -> IO ExitCode +systemRNAz inputFilePath outputFilePath = system ("RNAz " ++ inputFilePath ++ " >" ++ outputFilePath) + +-- | Parse the input as RNAz datatype +genParseRNAz :: GenParser Char st RNAz +genParseRNAz = do char '\n' many1 (oneOf "# ") string "RNAz" @@ -54,7 +62,7 @@ newline _rnaZResults <- many1 (try parseRNAzResult) _rnaZConsensus <- parseRNAzConsensus - return $ RNAzOutput _version _sequences _columns _readingDirection _meanPairwiseIdentity _shannonEntropy _gcContent _meanSingleSequenceMFE _consensusMFE _energyContribution _covarianceContribution _combinationsPair _meanZScore _structureConservationIndex _backgroundModel _decisionModel _svmDecisionValue _svmRNAClassProbability _prediction _rnaZResults _rnaZConsensus + return $ RNAz _version _sequences _columns _readingDirection _meanPairwiseIdentity _shannonEntropy _gcContent _meanSingleSequenceMFE _consensusMFE _energyContribution _covarianceContribution _combinationsPair _meanZScore _structureConservationIndex _backgroundModel _decisionModel _svmDecisionValue _svmRNAClassProbability _prediction _rnaZResults _rnaZConsensus -- | Parse a RNAz field containing a Double parseRNAzDoubleField :: String -> GenParser Char st Double @@ -92,7 +100,7 @@ _header <- many1 (noneOf "\n") newline notFollowedBy (string (">consensus")) - _resultSequence <- many1 (oneOf "~-NATUGCatugc") + _resultSequence <- many1 (oneOf "~_-.NATUGCatugc") newline _dotBracket <- many1 (oneOf "-().,") space @@ -118,7 +126,7 @@ parseRNAzConsensus = do string (">consensus") newline - _consensusSequence <- many1 (oneOf "~_-NATUGCatugc") + _consensusSequence <- many1 (oneOf "~_-.NATUGCatugc") newline _dotBracket <- many1 (oneOf "().,") space @@ -140,10 +148,12 @@ eof return $ RNAzConsensus _consensusSequence _dotBracket --- | parse RNAzOutput from input string -parseRNAz :: [Char] -> Either ParseError RNAzOutput -parseRNAz input = parse parseRNAzOutput "parseRNAzOutput" input +-- | parse RNAz from input string +parseRNAz :: [Char] -> Either ParseError RNAz +parseRNAz input = parse genParseRNAz "parseRNAz" input -- | parse from input filePath -readRNAz :: String -> IO (Either ParseError RNAzOutput) -readRNAz filePath = parseFromFile parseRNAzOutput filePath +readRNAz :: String -> IO (Either ParseError RNAz) +readRNAz filePath = do + parsedFile <- parseFromFile genParseRNAz filePath + CE.evaluate parsedFile diff --git a/src/Bio/ViennaRNAParser.hs b/src/Bio/ViennaRNAParser.hs --- a/src/Bio/ViennaRNAParser.hs +++ b/src/Bio/ViennaRNAParser.hs @@ -3,7 +3,9 @@ -- For more information on the Vienna RNAPackage consult: -- or the homepage of the current maintainer: -module Bio.ViennaRNAParser ( +module Bio.ViennaRNAParser ( + module Bio.RNAcofoldData, + module Bio.RNAcofoldParser, module Bio.RNAplexData, module Bio.RNAplexParser, module Bio.RNAzData, @@ -17,6 +19,8 @@ module Bio.RNAupData, module Bio.RNAupParser ) where +import Bio.RNAcofoldData +import Bio.RNAcofoldParser import Bio.RNAplexData import Bio.RNAplexParser import Bio.RNAzData diff --git a/test-suite/Spec.hs b/test-suite/Spec.hs new file mode 100644 --- /dev/null +++ b/test-suite/Spec.hs @@ -0,0 +1,2 @@ +-- test-suite/Spec.hs +{-# OPTIONS_GHC -F -pgmF hspec-discover #-}