diff --git a/ClustalParser.cabal b/ClustalParser.cabal --- a/ClustalParser.cabal +++ b/ClustalParser.cabal @@ -5,7 +5,7 @@ -- PVP summary: +-+------- breaking API changes -- | | +----- non-breaking API additions -- | | | +--- code changes with no API change -version: 1.0.0 +version: 1.0.1 synopsis: Libary for parsing Clustal tools output description: Currently contains parsers and datatypes for: clustalw2, clustalo . @@ -30,8 +30,8 @@ source-repository this type: git - location: https://github.com/eggzilla/ClustalParser/tree/1.0.0 - tag: 1.0.0 + location: https://github.com/eggzilla/ClustalParser/tree/1.0.1 + tag: 1.0.1 library -- Modules exported by the library. @@ -40,10 +40,14 @@ -- Other library packages from which modules are imported. build-depends: base >=4.5 && <5, parsec - + -- compiler options + ghc-options: -Wall -O2 -fno-warn-unused-do-bind -- Directories containing source files. hs-source-dirs: src executable ClustalParserTest main-is: ClustalParserTest.hs build-depends: base >= 4 && <= 5, cmdargs, ClustalParser + -- compiler options + ghc-options: -Wall -O2 + diff --git a/ClustalParserTest.hs b/ClustalParserTest.hs --- a/ClustalParserTest.hs +++ b/ClustalParserTest.hs @@ -1,11 +1,9 @@ module Main where import System.Environment (getArgs) -import System.IO -import System.Environment import Bio.ClustalParser -import Data.Either - + +main :: IO () main = do args <- getArgs let input_file = (head args) diff --git a/src/Bio/ClustalData.hs b/src/Bio/ClustalData.hs --- a/src/Bio/ClustalData.hs +++ b/src/Bio/ClustalData.hs @@ -42,7 +42,6 @@ } deriving (Show, Eq) - -- | Data structure for Clustal alignment format data ClustalAlignment = ClustalAlignment { @@ -73,16 +72,26 @@ } deriving (Show, Eq) -data ClustalGuideTree = ClustalGuideTree +-- | Data structure for structural Clustal alignment format +data StructuralClustalAlignment = StructuralClustalAlignment { - guideTreeEntries :: [ClustalGuideTreeEntry] + structuralAlignmentEntries :: [ClustalAlignmentEntry], + secondaryStructureTrack :: String, + energy :: Double } deriving (Show, Eq) -data ClustalGuideTreeEntry = ClustalGuideTreeEntry - { - entryDesignation :: String, - entryDistance :: Double +data StructuralClustalAlignmentSlice = StructuralClustalAlignmentSlice + { + structuralEntrySlices :: [StructuralClustalAlignmentEntrySlice] } deriving (Show, Eq) + +data StructuralClustalAlignmentEntrySlice = StructuralClustalAlignmentEntrySlice + { + structuralEntrySequenceSliceIdentifier :: String, + structuralEntryAlignedSliceSequence :: String + } + deriving (Show, Eq) + diff --git a/src/Bio/ClustalParser.hs b/src/Bio/ClustalParser.hs --- a/src/Bio/ClustalParser.hs +++ b/src/Bio/ClustalParser.hs @@ -1,17 +1,17 @@ -- | Parse Clustal output - +-- For more information on Clustal tools consult: module Bio.ClustalParser ( parseClustalAlignment, readClustalAlignment, + parseStructuralClustalAlignment, + readStructuralClustalAlignment, parseClustalSummary, readClustalSummary, module Bio.ClustalData ) where import Bio.ClustalData -import Text.ParserCombinators.Parsec -import Text.ParserCombinators.Parsec.Token -import Text.ParserCombinators.Parsec.Language (emptyDef) +import Text.ParserCombinators.Parsec import Control.Monad import Data.List @@ -35,7 +35,7 @@ newline newline string "Sequence format is " - sequenceFormat <- many1 (noneOf "\n") + sequenceFormat' <- many1 (noneOf "\n") newline sequenceParametersList <- many1 (try genParserSequenceParameters) string "Start of Pairwise alignments" @@ -45,7 +45,7 @@ newline pairwiseAlignmentSummaryList <- many1 genParserPairwiseAlignmentSummary string "Guide tree file created: [" - guideTreeFileName <- many1 (noneOf "]") + guideTreeFileName' <- many1 (noneOf "]") char ']' newline newline @@ -60,16 +60,16 @@ newline groupSummaryList <- many1 genParserGroupSummary string "Alignment Score " - alignmentScore <- many1 digit + alignmentScore' <- many1 digit newline newline string "CLUSTAL-Alignment file created [" - alignmentFileName <- many1 (noneOf "]") + alignmentFileName' <- many1 (noneOf "]") char ']' newline newline eof - return $ ClustalSummary version sequenceFormat sequenceParametersList pairwiseAlignmentSummaryList guideTreeFileName (readInt numberOfGroups) groupSummaryList (readInt alignmentScore) alignmentFileName + return $ ClustalSummary version sequenceFormat' sequenceParametersList pairwiseAlignmentSummaryList guideTreeFileName' (readInt numberOfGroups) groupSummaryList (readInt alignmentScore') alignmentFileName' genParserGroupSummary :: GenParser Char st GroupSummary genParserGroupSummary = do @@ -82,9 +82,9 @@ sequenceNumber <- optionMaybe (many1 digit) optional (many1 space) string "Score:" <|> string "Delayed" - groupScore <- optionMaybe (many1 digit) + groupScore' <- optionMaybe (many1 digit) newline - return $ GroupSummary (readInt groupIndex) (liftM readInt sequenceNumber) (liftM readInt groupScore) + return $ GroupSummary (readInt groupIndex) (liftM readInt sequenceNumber) (liftM readInt groupScore') genParserPairwiseAlignmentSummary :: GenParser Char st PairwiseAlignmentSummary genParserPairwiseAlignmentSummary = do @@ -139,14 +139,14 @@ genParserClustalAlignmentSlice :: GenParser Char st ClustalAlignmentSlice genParserClustalAlignmentSlice = do - entrySlices <- many1 genParserClustalEntrySlice + entrySlices' <- many1 genParserClustalEntrySlice --extract length of identifier and spacer to determine offset of conservation track - let offsetLenght = length (entrySequenceSliceIdentifier (head entrySlices)) + spacerLength (head entrySlices) + let offsetLenght = length (entrySequenceSliceIdentifier (head entrySlices')) + spacerLength (head entrySlices') spacerAndConservationTrackSlice <- many1 (noneOf "\n") - let conservationTrackSlice = drop offsetLenght spacerAndConservationTrackSlice + let conservationTrackSlice' = drop offsetLenght spacerAndConservationTrackSlice newline optional newline - return $ ClustalAlignmentSlice entrySlices conservationTrackSlice + return $ ClustalAlignmentSlice entrySlices' conservationTrackSlice' genParserClustalEntrySlice :: GenParser Char st ClustalAlignmentEntrySlice genParserClustalEntrySlice = do @@ -156,6 +156,87 @@ newline return $ ClustalAlignmentEntrySlice sliceIdentifier sliceSequence (length spacer) +--Structural Clustal Parser functions + +-- | Parse the input as ClustalAlignment datatype as used in mlocarna +genParserStructuralClustalAlignment :: GenParser Char st StructuralClustalAlignment +genParserStructuralClustalAlignment = do + genParseMlocarnaHeader + alignmentSlices <- many1 (try genParserStructuralClustalAlignmentSlice) + secondaryStructure <- genSecondaryStructure + energy' <- genParseEnergy + eof + return (mergeStructuralAlignmentSlices alignmentSlices secondaryStructure energy') + +genSecondaryStructure :: GenParser Char st String +genSecondaryStructure = do + string "alifold" + many1 space + secondaryStructure <- many1 (oneOf ".()") + space + return secondaryStructure + +genParseEnergy :: GenParser Char st Double +genParseEnergy = do + string "(" + many space + energy' <- many1 (noneOf " ") + optional space + char ('=') + many1 (noneOf "\n") + newline + return (readDouble energy') + +genParseMlocarnaHeader :: GenParser Char st String +genParseMlocarnaHeader = do + string "mLo" + many1 (noneOf "\n") + newline + string "Copyright" + many1 (noneOf "\n") + newline + newline + many1 genParseAlignmentProcessStep + newline + newline + return "" + +genParseAlignmentProcessStep :: GenParser Char st String +genParseAlignmentProcessStep = do + many1 (noneOf ".\n") + choice [try (string ("... ")), try (string ("..."))] + newline + return "" + +mergeStructuralAlignmentSlices :: [StructuralClustalAlignmentSlice] -> String -> Double -> StructuralClustalAlignment +mergeStructuralAlignmentSlices slices secondaryStructure energy' = alignment + where entrySlicesList = map structuralEntrySlices slices -- list of lists of entry slices + sequenceIdentifiers = map structuralEntrySequenceSliceIdentifier (head entrySlicesList) + alignmentEntriesListBySlice = map (map structuralEntryAlignedSliceSequence) entrySlicesList + transposedAlignmentEntriesListbySlice = transpose alignmentEntriesListBySlice + mergedAlignmentSequenceEntries = map concat transposedAlignmentEntriesListbySlice + mergedAlignmentEntries = map constructStructuralAlignmentEntries (zip sequenceIdentifiers mergedAlignmentSequenceEntries) + alignment = StructuralClustalAlignment mergedAlignmentEntries secondaryStructure energy' + +constructStructuralAlignmentEntries :: (String, String) -> ClustalAlignmentEntry +constructStructuralAlignmentEntries (entryIdentifier,entrySequence) = ClustalAlignmentEntry entryIdentifier entrySequence + +genParserStructuralClustalAlignmentSlice :: GenParser Char st StructuralClustalAlignmentSlice +genParserStructuralClustalAlignmentSlice = do + entrySlices' <- many1 (try genParserStructuralClustalEntrySlice) + optional newline + return $ StructuralClustalAlignmentSlice entrySlices' + +genParserStructuralClustalEntrySlice :: GenParser Char st StructuralClustalAlignmentEntrySlice +genParserStructuralClustalEntrySlice = do + sliceIdentifier <- many1 (noneOf " ") + many1 (char ' ') + sliceSequence <- many1 (oneOf "UAGCT-") + newline + return $ StructuralClustalAlignmentEntrySlice sliceIdentifier sliceSequence + +-- exported functions + -- | Parse Clustal alignment (.aln) from String parseClustalAlignment :: String -> Either ParseError ClustalAlignment parseClustalAlignment = parse genParserClustalAlignment "genParserClustalAlignment" @@ -163,6 +244,14 @@ -- | Parse Clustal alignment (.aln) from filehandle readClustalAlignment :: String -> IO (Either ParseError ClustalAlignment) readClustalAlignment = parseFromFile genParserClustalAlignment + +-- | Parse Clustal alignment (.aln) with secondary structure in dot-bracket notation from String (as produced by mlocarna) +parseStructuralClustalAlignment :: String -> Either ParseError StructuralClustalAlignment +parseStructuralClustalAlignment = parse genParserStructuralClustalAlignment "genParserClustalAlignment" + +-- | Parse Clustal alignment (.aln) with secondary structure in dot-bracket notation from filehandle (as produced by mlocarna) +readStructuralClustalAlignment :: String -> IO (Either ParseError StructuralClustalAlignment) +readStructuralClustalAlignment = parseFromFile genParserStructuralClustalAlignment -- | Parse Clustal summary (printed to STDOUT) from String parseClustalSummary :: String -> Either ParseError ClustalSummary