diff --git a/Biobase/GTF.hs b/Biobase/GTF.hs new file mode 100644 --- /dev/null +++ b/Biobase/GTF.hs @@ -0,0 +1,12 @@ +-- | Types and functions for NCBI BLAST+ +-- + +module Biobase.GTF + ( module Biobase.GTF.Types + , module Biobase.GTF.Import + ) where + +import Biobase.GTF.Import (gtfFromFile) +import Biobase.GTF.Export +import Biobase.GTF.Types + diff --git a/Biobase/GTF/Export.hs b/Biobase/GTF/Export.hs new file mode 100644 --- /dev/null +++ b/Biobase/GTF/Export.hs @@ -0,0 +1,18 @@ + +-- | + +module Biobase.GTF.Export where +import Biobase.GTF.Types +import qualified Data.ByteString.Lazy.Char8 as B +import qualified Data.Vector as V +import Data.List + +instance Show GTF where + show (GTF _Entries _Sequence) + | not (null _Entries) = entriesString + | otherwise = "" + where entriesString = concatMap show _Entries + +instance Show GTFEntry where + show (GTFEntry _gtfSeqid _gtfSource _gtfType _gtfStart _gtfEnd _gtfScore _gtfStrand _gtfPhase _gtfAttributes) = + (B.unpack _gtfSeqid) ++ "\t" ++ (B.unpack _gtfSource) ++ "\t" ++ (B.unpack _gtfType) ++ "\t" ++ show _gtfStart ++ "\t" ++ show _gtfEnd ++ "\t" ++ (B.unpack _gtfScore) ++ "\t" ++ [_gtfStrand] ++ "\t" ++ (B.unpack _gtfPhase) ++ "\t" ++ (intercalate ";" (V.toList (V.map B.unpack _gtfAttributes))) ++ "\n" diff --git a/Biobase/GTF/Import.hs b/Biobase/GTF/Import.hs new file mode 100644 --- /dev/null +++ b/Biobase/GTF/Import.hs @@ -0,0 +1,89 @@ +{-# LANGUAGE OverloadedStrings #-} +{-# LANGUAGE LambdaCase #-} + +-- | Parses GTF + +module Biobase.GTF.Import (gtfFromFile, + parseGTFs, + ) where + +import Prelude hiding (takeWhile) +import Data.Attoparsec.ByteString.Char8 hiding (isSpace) +import qualified Data.Attoparsec.ByteString.Lazy as L +import qualified Data.ByteString.Char8 as C +import qualified Data.ByteString.Builder as S +import qualified Data.ByteString.Lazy.Char8 as B +import qualified Data.Vector as V +import System.Directory +import Data.Char +import Control.Monad +import Debug.Trace +import Text.Printf +import Biobase.GTF.Types +import qualified Data.Word8 as W + +-- | reads and parses GTF from provided filePath +gtfFromFile :: String -> IO [GTF] +gtfFromFile filePath = do + printf "# reading GTF from file %s\n" filePath + fileExists <- doesFileExist filePath + if fileExists + then parseGTFs <$> B.readFile filePath + else fail ("# GTF file \"%s\" does not exist\n" ++ filePath) + +-- | Read a lazy bytestring and stream out a list of @GTFs@'s. +-- In case, there is a parse error "late" in the file, we might have +-- already streamed out some (or many!) of these results. + +parseGTFs :: B.ByteString -> [GTF] +parseGTFs = go + where go xs = case L.parse genParseGTF xs of + L.Fail remainingInput ctxts err -> error $ "parseGTFs failed! " ++ err ++ " ctxt: " ++ show ctxts ++ " head of remaining input: " ++ B.unpack (B.take 1000 remainingInput) + L.Done remainingInput btr + | B.null remainingInput -> [btr] + | otherwise -> btr : go remainingInput + +genParseGTF :: Parser GTF +genParseGTF = do + skipMany (try genParseGTFComment) + _entry <- many1 (try genParseGTFEntry) "GTF entry" + return $ GTF (V.fromList _entry) B.empty + +genParseGTFComment :: Parser String +genParseGTFComment = do + string "#" + takeWhile1 (/= '\n') + endOfLine + return $ "" + +genParseGTFEntry :: Parser GTFEntry +genParseGTFEntry = do + _gtfSeqid <- takeWhile1 (/= '\t') "seqid" + char '\t' + _gtfSource <- takeWhile1 (/= '\t') "source" + char '\t' + _gtfType <- takeWhile1 (/= '\t') "type" + char '\t' + _gtfStart <- decimal "start" + char '\t' + _gtfEnd <- decimal "end" + char '\t' + _gtfScore <- takeWhile1 (/= '\t') "score" + char '\t' + _gtfStrand <- (choice [char '+', char '-', char '.']) "strand" + char '\t' + _gtfPhase <- takeWhile1 (/= '\t') "phase" + char '\t' + _gtfAttributes <- genParseGTFAttributes "GTF attributes" + endOfLine + skipMany (try genParseGTFComment) + return $ GTFEntry (B.fromStrict _gtfSeqid) (B.fromStrict _gtfSource) (B.fromStrict _gtfType) _gtfStart _gtfEnd (B.fromStrict _gtfScore) _gtfStrand (B.fromStrict _gtfPhase) (V.fromList _gtfAttributes) + +genParseGTFAttributes :: Parser [B.ByteString] +genParseGTFAttributes = do + _gtfAtributesString <- takeTill (\a -> a == '\n') "attributes" + let _gtfAtributes = map B.fromStrict (C.split ';' _gtfAtributesString) + return $ _gtfAtributes + +toLB :: C.ByteString -> B.ByteString +toLB = S.toLazyByteString . S.byteString diff --git a/Biobase/GTF/Types.hs b/Biobase/GTF/Types.hs new file mode 100644 --- /dev/null +++ b/Biobase/GTF/Types.hs @@ -0,0 +1,38 @@ + +-- | Encoding of GTF + +module Biobase.GTF.Types where + +import Prelude hiding (takeWhile) +import Data.Attoparsec.ByteString.Char8 hiding (isSpace) +import qualified Data.Attoparsec.ByteString.Lazy as L +import qualified Data.ByteString.Char8 as C +import qualified Data.ByteString.Builder as S +import qualified Data.ByteString.Lazy.Char8 as B +import qualified Data.Vector as V +import System.Directory +import Data.Char +import Control.Monad +import Debug.Trace +import Text.Printf + + -- | Datastructure for GTF http://gmod.org/wiki/GTF +data GTF = GTF + { gtfEntries :: !(V.Vector GTFEntry), + gtfSequence :: !B.ByteString + } + deriving (Eq) + +-- | Datastructure for data lines of GTF http://gmod.org/wiki/GTF +data GTFEntry = GTFEntry + { gtfSeqid :: !B.ByteString, + gtfSource :: !B.ByteString, + gtfType :: !B.ByteString, + gtfStart :: Int, + gtfEnd :: Int, + gtfScore :: !B.ByteString, + gtfStrand :: Char, + gtfPhase :: B.ByteString, + gtfAttributes :: !(V.Vector B.ByteString) + } + deriving (Eq) diff --git a/BiobaseEnsembl.cabal b/BiobaseEnsembl.cabal --- a/BiobaseEnsembl.cabal +++ b/BiobaseEnsembl.cabal @@ -1,5 +1,5 @@ name: BiobaseEnsembl -version: 0.2.0.0 +version: 0.2.0.1 author: Florian Eggenhofer maintainer: egg@informatik.uni-freiburg.de homepage: https://github.com/eggzilla/BiobaseEnsembl @@ -11,7 +11,7 @@ build-type: Simple stability: experimental cabal-version: >= 1.10.0 -tested-with: GHC == 8.0.2, GHC == 8.2.2, GHC == 8.4.4, GHC == 8.6.5, GHC == 8.8.1 +tested-with: GHC == 8.6.5, GHC == 8.8.1 synopsis: Ensembl related datastructures and functions description: This library contains high through put sequencing and Ensembl-related functionality: @@ -53,6 +53,10 @@ Biobase.GFF3.Import Biobase.GFF3.Export Biobase.GFF3.Types + Biobase.GTF + Biobase.GTF.Import + Biobase.GTF.Export + Biobase.GTF.Types Biobase.Ensembl.REST.Types ghc-options: @@ -64,5 +68,5 @@ source-repository this type: git - location: https://github.com/eggzilla/BiobaseEnsembl/tree/0.2.0.0 - tag: 0.2.0.0 + location: https://github.com/eggzilla/BiobaseEnsembl/tree/0.2.0.1 + tag: 0.2.0.1 diff --git a/ChangeLog.md b/ChangeLog.md --- a/ChangeLog.md +++ b/ChangeLog.md @@ -1,9 +1,5 @@ -*-change-log-*- -### 0.2.0.0 [Florian Eggenhofer](mailto:egg@cs.uni-freiburg.de) 17. December 2019 - - * Construction start from input alignment for Scan and Alien - * Alien is working fully offline, by using offline taxonomy database ### 0.1.0.0 [Florian Eggenhofer](mailto:egg@cs.uni-freiburg.de) 22. August 2018