diff --git a/RNAlien.cabal b/RNAlien.cabal --- a/RNAlien.cabal +++ b/RNAlien.cabal @@ -1,5 +1,5 @@ name: RNAlien -version: 1.3.0 +version: 1.3.1 synopsis: Unsupervized construction of RNA family models description: RNAlien is a tool for automatic construction of RNAfamily models from a single sequence. . @@ -50,8 +50,8 @@ source-repository this type: git - location: https://github.com/eggzilla/RNAlien/tree/1.3.0 - tag: 1.3.0 + location: https://github.com/eggzilla/RNAlien/tree/1.3.1 + tag: 1.3.1 executable RNAlien Hs-Source-Dirs: ./src/Bio/ @@ -86,7 +86,7 @@ Library Hs-Source-Dirs: ./src/ ghc-options: -Wall -O2 -fno-warn-unused-do-bind - build-depends: base >=4.5 && <5, cmdargs, ViennaRNAParser>=1.3.2, process, directory, blastxml>=0.3.2, biofasta, parsec, random, BlastHTTP>=1.2.1, biocore, bytestring, Taxonomy >= 1.0.2, either-unwrap, containers, ClustalParser>=1.2.0, EntrezHTTP>=1.0.3, vector, edit-distance, cassava, matrix, hierarchical-clustering, filepath, HTTP, http-conduit, hxt, network, aeson, text, transformers, pureMD5, http-types, text-metrics + build-depends: base >=4.5 && <5, cmdargs, ViennaRNAParser>=1.3.2, process, directory, blastxml>=0.3.2, biofasta, parsec, random, BlastHTTP>=1.2.1, biocore, bytestring, Taxonomy >= 1.0.2, either-unwrap, containers, ClustalParser>=1.2.1, EntrezHTTP>=1.0.3, vector, edit-distance, cassava, matrix, hierarchical-clustering, filepath, HTTP, http-conduit, hxt, network, aeson, text, transformers, pureMD5, http-types, text-metrics Exposed-Modules: Bio.RNAlienData Bio.RNAlienLibrary Bio.RNAcentralHTTP diff --git a/src/Bio/RNAlien.hs b/src/Bio/RNAlien.hs --- a/src/Bio/RNAlien.hs +++ b/src/Bio/RNAlien.hs @@ -85,7 +85,7 @@ logMessage "Error: Input fasta file is empty.\n" temporaryDirectoryPath else do let iterationNumber = 0 - let tools = ["mlocarna","RNAfold","RNAalifold","cmcalibrate","cmstat","cmbuild","RNAz","RNAcode"] + let tools = if inputQuerySelectionMethod == "clustering" then ["clustalo","mlocarna","RNAfold","RNAalifold","cmcalibrate","cmstat","cmbuild","RNAz","RNAcode"] else ["mlocarna","RNAfold","RNAalifold","cmcalibrate","cmstat","cmbuild","RNAz","RNAcode"] toolsCheck <- checkTools tools inputQuerySelectionMethod temporaryDirectoryPath -- Check required commandline tools if isLeft toolsCheck @@ -93,7 +93,7 @@ putStrLn ("Error - Not all required tools could be found in $PATH: " ++ fromLeft toolsCheck ++ "\n") logMessage ("Error - Not all required tools could be found in $PATH: " ++ fromLeft toolsCheck ++ "\n") temporaryDirectoryPath else do - logToolVersions temporaryDirectoryPath + logToolVersions inputQuerySelectionMethod temporaryDirectoryPath let inputSequence = reformatFasta (head inputFasta) initialTaxId <- setInitialTaxId inputBlastDatabase temporaryDirectoryPath inputTaxId inputSequence let checkedTaxonomyRestriction = checkTaxonomyRestriction taxonomyRestriction diff --git a/src/Bio/RNAlienLibrary.hs b/src/Bio/RNAlienLibrary.hs --- a/src/Bio/RNAlienLibrary.hs +++ b/src/Bio/RNAlienLibrary.hs @@ -851,13 +851,13 @@ -- | Compute identity of sequences -stringIdentity :: String -> String -> Double -stringIdentity string1 string2 = identityPercent - where distance = ED.levenshteinDistance costs string1 string2 - --Replication of RNAz select sequences requires only allowing substitutions - costs = ED.defaultEditCosts {ED.deletionCosts = ED.ConstantCost 100,ED.insertionCosts = ED.ConstantCost 100,ED.transpositionCosts = ED.ConstantCost 100} - maximumDistance = maximum [length string1,length string2] - identityPercent = 1 - (fromIntegral distance/fromIntegral maximumDistance) +-- stringIdentity :: String -> String -> Double +-- stringIdentity string1 string2 = identityPercent +-- where distance = ED.levenshteinDistance costs string1 string2 +-- --Replication of RNAz select sequences requires only allowing substitutions +-- costs = ED.defaultEditCosts {ED.deletionCosts = ED.ConstantCost 100,ED.insertionCosts = ED.ConstantCost 100,ED.transpositionCosts = ED.ConstantCost 100} +-- maximumDistance = maximum [length string1,length string2] +-- identityPercent = 1 - (fromIntegral distance/fromIntegral maximumDistance) -- | Compute identity of sequences sequenceIdentity :: Sequence -> Sequence -> Double @@ -1466,26 +1466,33 @@ logMessage ("Tools : " ++ intercalate "," tools ++ "\n") temporaryDirectoryPath return (Right "Tools ok") -logToolVersions :: String -> IO () -logToolVersions temporaryDirectoryPath = do +logToolVersions :: String -> String -> IO () +logToolVersions inputQuerySelectionMethod temporaryDirectoryPath = do let clustaloversionpath = temporaryDirectoryPath ++ "log/clustalo.version" let mlocarnaversionpath = temporaryDirectoryPath ++ "log/mlocarna.version" let rnafoldversionpath = temporaryDirectoryPath ++ "log/RNAfold.version" let infernalversionpath = temporaryDirectoryPath ++ "log/Infernal.version" - _ <- system ("clustalo --version >" ++ clustaloversionpath) + --_ <- system ("clustalo --version >" ++ clustaloversionpath) _ <- system ("mlocarna --version >" ++ mlocarnaversionpath) _ <- system ("RNAfold --version >" ++ rnafoldversionpath) _ <- system ("cmcalibrate -h >" ++ infernalversionpath) -- _ <- system ("RNAz" ++ rnazversionpath) -- _ <- system ("CMCompare >" ++ infernalversionpath) - clustaloversion <- readFile clustaloversionpath mlocarnaversion <- readFile mlocarnaversionpath rnafoldversion <- readFile rnafoldversionpath infernalversionOutput <- readFile infernalversionpath let infernalversion = lines infernalversionOutput !! 1 - let messageString = "Clustalo version: " ++ clustaloversion ++ "mlocarna version: " ++ mlocarnaversion ++ "RNAfold version: " ++ rnafoldversion ++ "infernalversion: " ++ infernalversion ++ "\n" - logMessage messageString temporaryDirectoryPath + if inputQuerySelectionMethod == "clustering" + then do + _ <- system ("clustalo --version >" ++ clustaloversionpath) + clustaloversion <- readFile clustaloversionpath + let messageString = "Clustalo version: " ++ clustaloversion ++ "mlocarna version: " ++ mlocarnaversion ++ "RNAfold version: " ++ rnafoldversion ++ "infernalversion: " ++ infernalversion ++ "\n" + logMessage messageString temporaryDirectoryPath + else do + let messageString = "mlocarna version: " ++ mlocarnaversion ++ "RNAfold version: " ++ rnafoldversion ++ "infernalversion: " ++ infernalversion ++ "\n" + logMessage messageString temporaryDirectoryPath + checkTool :: String -> IO (Either String String) checkTool tool = do toolcheck <- findExecutable tool @@ -1526,7 +1533,7 @@ let reformatedClustalPath = evaluationDirectoryFilepath ++ "result.clustal.reformated" let cmFilepath = tempDirPath staticOptions ++ "result.cm" let resultSequences = (inputFasta mCResult):map nucleotideSequence (concatMap sequenceRecords (taxRecords mCResult)) - let resultNumber = length resultSequences + 1 + let resultNumber = length resultSequences let rnaCentralQueries = map RCH.buildSequenceViaMD5Query resultSequences rnaCentralEntries <- RCH.getRNACentralEntries rnaCentralQueries let rnaCentralEvaluationResult = RCH.showRNAcentralAlienEvaluation rnaCentralEntries @@ -1647,9 +1654,9 @@ return (Right (formatedIdMatrix,selectedClustalpath)) else return (Left (show (fromLeft parsedClustalInput))) else do - let clustalTextLines = T.lines clustalText - let headerClustalTextLines = T.unlines (take 2 clustalTextLines) - let headerlessClustalTextLines = T.unlines (drop 2 clustalTextLines) + let clustalLines = T.lines clustalText + let headerClustalTextLines = T.unlines (take 2 clustalLines) + let headerlessClustalTextLines = T.unlines (drop 2 clustalLines) let reformatedClustalText = T.map reformatRNACodeAln headerlessClustalTextLines TI.writeFile selectedClustalpath (headerClustalTextLines `T.append` (T.singleton '\n') `T.append` reformatedClustalText) return (Right ([],clustalFilepath))