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: 1.2.0 +version: 1.2.1 synopsis: Libary for parsing Clustal tools output description: Currently contains parsers and datatypes for: clustalw2, clustalo, mlocarna, cmalign . @@ -18,7 +18,7 @@ Cmalign is a multiple sequence alignment program based on RNA family models and produces ,among others, clustal output. It is part of infernal . -license: GPL +license: GPL-3.0 license-file: LICENSE author: Florian Eggenhofer maintainer: egg@tbi.univie.ac.at @@ -36,8 +36,8 @@ source-repository this type: git - location: https://github.com/eggzilla/ClustalParser/tree/1.2.0 - tag: 1.2.0 + location: https://github.com/eggzilla/ClustalParser/tree/1.2.1 + tag: 1.2.1 library -- Modules exported by the library. @@ -58,7 +58,7 @@ ghc-options: -Wall -O2 test-suite hspec - build-depends: base, parsec, ClustalParser, hspec >= 1.8 + build-depends: base, parsec, ClustalParser, hspec >= 1.8, text hs-source-dirs: test-suite main-is: Spec.hs type: exitcode-stdio-1.0 diff --git a/ClustalParserTest.hs b/ClustalParserTest.hs --- a/ClustalParserTest.hs +++ b/ClustalParserTest.hs @@ -1,7 +1,9 @@ +--runghc -package-db --ghc-arg=.cabal-sandbox/x86_64-linux-ghc-8.0.1-packages.conf.d ClustalParserTest.hs struct-multiline.mlocarna module Main where import System.Environment (getArgs) import Bio.ClustalParser +--import Data.Either.Unwrap main :: IO () main = do @@ -10,4 +12,5 @@ -- read Clustal outputfile parsedinput <- readStructuralClustalAlignment input_file print parsedinput + --print $ structuralAlignmentEntries (fromRight parsedinput) diff --git a/changelog b/changelog --- a/changelog +++ b/changelog @@ -1,4 +1,6 @@ -*-change-log-*- +1.2.1 Florian Eggenhofer 06. February 2017 + * Structural alignment parser now works with multiline consensus structures 1.2.0 Florian Eggenhofer 07. January 2017 * Changed datastructures for sequence identifers and sequences to Data.Text 1.1.4 Florian Eggenhofer 30. May 2016 diff --git a/src/Bio/ClustalData.hs b/src/Bio/ClustalData.hs --- a/src/Bio/ClustalData.hs +++ b/src/Bio/ClustalData.hs @@ -66,7 +66,7 @@ | totalSequenceLength == 0 = [] | currentWindowPosition < totalSequenceLength = showAlignmentBlock longestSequenceIdLength currentWindowPosition _alignmentEntries _conservationTrack ++ (showAlignment totalSequenceLength longestSequenceIdLength (currentWindowPosition + 60) _alignmentEntries _conservationTrack) | currentWindowPosition == totalSequenceLength = [] - | otherwise = "" + | otherwise = "" showAlignmentBlock :: Int -> Int -> [ClustalAlignmentEntry] -> T.Text -> String showAlignmentBlock longestSequenceIdLength currentWindowPosition _alignmentEntries _conservationTrack = blockString @@ -92,12 +92,12 @@ deriving (Show, Eq) data ClustalAlignmentEntrySlice = ClustalAlignmentEntrySlice - { - entrySequenceSliceIdentifier :: String, - entryAlignedSliceSequence :: String, - spacerLength :: Int - } - deriving (Show, Eq) + { + entrySequenceSliceIdentifier :: String, + entryAlignedSliceSequence :: String, + spacerLength :: Int + } + deriving (Show, Eq) -- | Data structure for structural Clustal alignment format data StructuralClustalAlignment = StructuralClustalAlignment @@ -117,11 +117,11 @@ totalSequenceLength = T.length (entryAlignedSequence (head _alignmentEntries)) alignmentString = showAlignment totalSequenceLength longestSequenceIdLength 0 _alignmentEntries _secondaryStructureTrack -data StructuralClustalAlignmentSlice = StructuralClustalAlignmentSlice - { - structuralEntrySlices :: [StructuralClustalAlignmentEntrySlice] - } - deriving (Show, Eq) +-- data StructuralClustalAlignmentSlice = StructuralClustalAlignmentSlice +-- { +-- structuralEntrySlices :: [StructuralClustalAlignmentEntrySlice] +-- } +-- deriving (Show, Eq) data StructuralClustalAlignmentEntrySlice = StructuralClustalAlignmentEntrySlice { diff --git a/src/Bio/ClustalParser.hs b/src/Bio/ClustalParser.hs --- a/src/Bio/ClustalParser.hs +++ b/src/Bio/ClustalParser.hs @@ -15,7 +15,6 @@ import Control.Monad import Data.List import qualified Data.Text as T - readDouble :: String -> Double readDouble = read @@ -124,18 +123,17 @@ mergealignmentSlices :: [ClustalAlignmentSlice] -> ClustalAlignment mergealignmentSlices slices = alignment - where entrySlicesList = map entrySlices slices -- list of lists of entry slices - sequenceIdentifiers = nub (map entrySequenceSliceIdentifier (head entrySlicesList)) - alignmentEntriesListBySlice = map (map entryAlignedSliceSequence) entrySlicesList - transposedAlignmentEntriesListbySlice = transpose alignmentEntriesListBySlice - mergedAlignmentSequenceEntries = map concat transposedAlignmentEntriesListbySlice - mergedAlignmentEntries = map constructAlignmentEntries (zip sequenceIdentifiers mergedAlignmentSequenceEntries) - conservationTrackSlices = map conservationTrackSlice slices - mergedConservationTrack = concat conservationTrackSlices - alignment = ClustalAlignment mergedAlignmentEntries (T.pack mergedConservationTrack) + where entrySlicesList = concatMap entrySlices slices -- list of lists of entry slices + sequenceIdentifiers = nub (map entrySequenceSliceIdentifier entrySlicesList) + mergedAlignmentEntries = map (constructAlignmentEntries entrySlicesList) sequenceIdentifiers + mergedConservationTrack = concatMap conservationTrackSlice slices + alignment = ClustalAlignment mergedAlignmentEntries (T.pack mergedConservationTrack) -constructAlignmentEntries :: (String, String) -> ClustalAlignmentEntry -constructAlignmentEntries (entryIdentifier,entrySequence) = ClustalAlignmentEntry (T.pack entryIdentifier) (T.pack entrySequence) +constructAlignmentEntries :: [ClustalAlignmentEntrySlice] -> String -> ClustalAlignmentEntry +constructAlignmentEntries slices entryIdentifier= entry + where currentSlices = filter (\a -> entrySequenceSliceIdentifier a == entryIdentifier) slices + entrySequence = concatMap entryAlignedSliceSequence currentSlices + entry = ClustalAlignmentEntry (T.pack entryIdentifier) (T.pack entrySequence) genParserClustalAlignmentSlice :: GenParser Char st ClustalAlignmentSlice genParserClustalAlignmentSlice = do @@ -166,16 +164,21 @@ secondaryStructure <- genSecondaryStructure energy' <- genParseEnergy eof - return (mergeStructuralAlignmentSlices alignmentSlices secondaryStructure energy') + return (mergeStructuralAlignmentSlices (concat alignmentSlices) secondaryStructure energy') genSecondaryStructure :: GenParser Char st String genSecondaryStructure = do string "alifold" - many1 space - secondaryStructure <- many1 (oneOf ".()") - space - return secondaryStructure + secondaryStructure <- many1 (try genSecondaryStructureSlice) + return (concat secondaryStructure) +genSecondaryStructureSlice :: GenParser Char st String +genSecondaryStructureSlice = do + many1 space + secondaryStructureSlice <- many1 (try (oneOf ".()")) + choice [try (string "\n"),try (string " ")] + return secondaryStructureSlice + genParseEnergy :: GenParser Char st Double genParseEnergy = do string "(" @@ -204,24 +207,23 @@ many1 newline return "" -mergeStructuralAlignmentSlices :: [StructuralClustalAlignmentSlice] -> String -> Double -> StructuralClustalAlignment +mergeStructuralAlignmentSlices :: [StructuralClustalAlignmentEntrySlice] -> 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 (T.pack secondaryStructure) energy' + where sequenceIdentifiers = nub (map structuralEntrySequenceSliceIdentifier slices) + mergedAlignmentEntries = map (constructStructuralAlignmentEntries slices) sequenceIdentifiers + alignment = StructuralClustalAlignment mergedAlignmentEntries (T.pack secondaryStructure) energy' -constructStructuralAlignmentEntries :: (String, String) -> ClustalAlignmentEntry -constructStructuralAlignmentEntries (entryIdentifier,entrySequence) = ClustalAlignmentEntry (T.pack entryIdentifier) (T.pack entrySequence) +constructStructuralAlignmentEntries :: [StructuralClustalAlignmentEntrySlice] -> String -> ClustalAlignmentEntry +constructStructuralAlignmentEntries slices entryIdentifier= entry + where currentSlices = filter (\a -> structuralEntrySequenceSliceIdentifier a == entryIdentifier) slices + entrySequence = concatMap structuralEntryAlignedSliceSequence currentSlices + entry = ClustalAlignmentEntry (T.pack entryIdentifier) (T.pack entrySequence) -genParserStructuralClustalAlignmentSlice :: GenParser Char st StructuralClustalAlignmentSlice +genParserStructuralClustalAlignmentSlice :: GenParser Char st [StructuralClustalAlignmentEntrySlice] genParserStructuralClustalAlignmentSlice = do entrySlices' <- many1 (try genParserStructuralClustalEntrySlice) optional newline - return $ StructuralClustalAlignmentSlice entrySlices' + return entrySlices' genParserStructuralClustalEntrySlice :: GenParser Char st StructuralClustalAlignmentEntrySlice genParserStructuralClustalEntrySlice = do @@ -238,23 +240,23 @@ parseClustalAlignment = parse genParserClustalAlignment "genParserClustalAlignment" -- | Parse Clustal alignment (.aln) from filehandle -readClustalAlignment :: String -> IO (Either ParseError ClustalAlignment) +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 :: String -> Either ParseError StructuralClustalAlignment parseStructuralClustalAlignment = parse genParserStructuralClustalAlignment "genParserStructuralClustalAlignment" -- | Parse Clustal alignment (.aln) with secondary structure in dot-bracket notation from filehandle (as produced by mlocarna) -readStructuralClustalAlignment :: String -> IO (Either ParseError StructuralClustalAlignment) +readStructuralClustalAlignment :: String -> IO (Either ParseError StructuralClustalAlignment) readStructuralClustalAlignment = parseFromFile genParserStructuralClustalAlignment -- | Parse Clustal summary (printed to STDOUT) from String -parseClustalSummary :: String -> Either ParseError ClustalSummary +parseClustalSummary :: String -> Either ParseError ClustalSummary parseClustalSummary = parse genParserClustalSummary "genParserClustalSummary" -- | Parse Clustal summary (printed to STDOUT) from file -readClustalSummary :: String -> IO (Either ParseError ClustalSummary) +readClustalSummary :: String -> IO (Either ParseError ClustalSummary) readClustalSummary = parseFromFile genParserClustalSummary -- | Parse nucleotide sequence. Allowed letters according to IUPAC