diff --git a/TaxonomyTools.cabal b/TaxonomyTools.cabal --- a/TaxonomyTools.cabal +++ b/TaxonomyTools.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: Tool for parsing, processing, comparing and visualizing taxonomy data description: Taxonomy Tool utilizes functions from the Taxonomy library to provide . @@ -53,32 +53,32 @@ source-repository this type: git - location: https://github.com/eggzilla/TaxonomyTools/tree/1.0.0 - tag: 1.0.0 + location: https://github.com/eggzilla/TaxonomyTools/tree/1.0.1 + tag: 1.0.1 executable TaxIds2Text Hs-Source-Dirs: ./src/Bio/ main-is: TaxIds2Text.hs ghc-options: -Wall -O2 - build-depends: base >=4.5 && <5, cmdargs, Taxonomy, either-unwrap, fgl, bytestring + build-depends: base >=4.5 && <5, cmdargs, Taxonomy>=1.0.2, either-unwrap, fgl, text executable Accessions2TaxIds Hs-Source-Dirs: ./src/Bio/ main-is: Accessions2TaxIds.hs ghc-options: -Wall -O2 -fno-warn-unused-do-bind - build-depends: base >=4.5 && <5, cmdargs, process, directory, Taxonomy, either-unwrap, fgl, cassava, vector, bytestring, parsec, EntrezHTTP, hxt + build-depends: base >=4.5 && <5, cmdargs, process, directory, Taxonomy>=1.0.2, either-unwrap, fgl, cassava>=0.4.5.1, vector, bytestring, parsec, EntrezHTTP>=1.0.2, hxt executable TaxIds2TreeCompare Hs-Source-Dirs: ./src/Bio/ main-is: TaxIds2TreeCompare.hs ghc-options: -Wall -O2 - build-depends: base >=4.5 && <5, cmdargs, process, directory, Taxonomy, either-unwrap, fgl, cassava, vector, bytestring + build-depends: base >=4.5 && <5, cmdargs, process, directory, Taxonomy>=1.0.2, either-unwrap, fgl, cassava>=0.4.5.1, vector, bytestring executable TaxIds2Tree Hs-Source-Dirs: ./src/Bio/ main-is: TaxIds2Tree.hs ghc-options: -Wall -O2 - build-depends: base >=4.5 && <5, cmdargs, process, directory, Taxonomy, either-unwrap, fgl, cassava, vector, bytestring, aeson + build-depends: base >=4.5 && <5, cmdargs, process, directory, Taxonomy>=1.0.2, either-unwrap, fgl, cassava>=0.4.5.1, vector, bytestring, aeson executable TaxIdsDistance Hs-Source-Dirs: ./src/Bio/ diff --git a/src/Bio/TaxIds2Text.hs b/src/Bio/TaxIds2Text.hs --- a/src/Bio/TaxIds2Text.hs +++ b/src/Bio/TaxIds2Text.hs @@ -7,7 +7,7 @@ import System.Console.CmdArgs import Data.Either.Unwrap -import qualified Data.ByteString.Char8 as BC +import qualified Data.Text.Lazy as TL import Bio.Taxonomy import Data.Maybe import Text.Read @@ -49,7 +49,7 @@ mapM_ putStrLn outputCSV printSimpleNode :: SimpleTaxon -> String -printSimpleNode snode = show (simpleRank snode) ++ "," ++ BC.unpack (simpleScientificName snode) +printSimpleNode snode = show (simpleRank snode) ++ "," ++ TL.unpack (simpleScientificName snode) readMaybeRank :: String -> Maybe Rank readMaybeRank inputString = readMaybe inputString diff --git a/src/Bio/TaxIds2Tree.hs b/src/Bio/TaxIds2Tree.hs --- a/src/Bio/TaxIds2Tree.hs +++ b/src/Bio/TaxIds2Tree.hs @@ -23,6 +23,7 @@ alienCSVFilePath ::String, levels :: Int, outputFormat :: String, + withRank :: Bool, outputDirectoryPath :: String } deriving (Show,Data,Typeable) @@ -33,8 +34,9 @@ alienCSVFilePath = def &= name "r" &= help "Path to RNAlienResult CSV. Alternative to input taxonomy id list", levels = (1 ::Int) &= name "l" &= help "Number defining maximum distance from root for nodes in subtree.", outputFormat = "dot" &= name "f" &= help "Requested output format (json,dot). Default: dot", + withRank = True &= name "w" &= help "Add taxonomic ranks to output. Default: True", outputDirectoryPath = def &= name "o" &= help "Path to output directory" - } &= summary "TaxIds2Tree - List of taxonomy ids is converted into a graphical tree representation either as .svg (via graphviz) or as .json (via d3js)" &= help "Florian Eggenhofer - 2015" &= verbosity + } &= summary "TaxIds2Tree - List of taxonomy ids is converted into a graphical tree representation either as .svg (via graphviz) or as .json (via d3js)" &= help "Florian Eggenhofer - 2016" &= verbosity main :: IO () main = do @@ -48,34 +50,36 @@ putStrLn "Provide a path to input taxonomy id list or to RNAlienResult CSV." else do -- input AlienCSV path present - taxidtableentries <- extractTaxidsAlienCSV alienCSVFilePath - let graph = fromRight graphOutput - let currentSubgraph = extractTaxonomySubTreebyLevel taxidtableentries graph (Just levels) - --let subdiagram = drawTaxonomy (grev currentSubgraph) - --writeFile (outputDirectoryPath ++ "taxonomy.dot") subdiagram - generateOutput outputFormat outputDirectoryPath currentSubgraph + decodedCsvOutput <- extractTaxidsAlienCSV alienCSVFilePath + if (isRight decodedCsvOutput) + then do + let decodedCsvList = (fromRight decodedCsvOutput) + let taxidtableentries = V.map firstOfTaxCSVTriple decodedCsvList + let graph = fromRight graphOutput + let currentSubgraph = extractTaxonomySubTreebyLevelNew (V.toList taxidtableentries) graph (Just levels) + generateOutput outputFormat outputDirectoryPath withRank currentSubgraph + else do + writeFile (outputDirectoryPath ++ "taxonomy.json") (show (fromLeft decodedCsvOutput)) else do -- input taxid path present taxidtable <- readFile taxNodeListFilePath let taxidtableentries = map (\l -> read l :: Int) (lines taxidtable) let graph = fromRight graphOutput - let currentSubgraph = extractTaxonomySubTreebyLevel taxidtableentries graph (Just levels) - --let subdiagram = drawTaxonomy (grev currentSubgraph) - --writeFile (outputDirectoryPath ++ "taxonomy.dot") subdiagram - generateOutput outputFormat outputDirectoryPath currentSubgraph + let currentSubgraph = extractTaxonomySubTreebyLevel taxidtableentries graph (Just levels) + writeTree outputFormat outputDirectoryPath withRank currentSubgraph -- | generate output -generateOutput :: String -> String -> Gr SimpleTaxon Double -> IO () -generateOutput requestedFormat outputDirectoryPath inputGraph = do +generateOutput :: String -> String -> Bool -> Gr SimpleTaxon Double -> IO () +generateOutput requestedFormat outputDirectoryPath withRank inputGraph = do case requestedFormat of - "dot" -> generateDotOutput outputDirectoryPath inputGraph + "dot" -> generateDotOutput outputDirectoryPath withRank inputGraph "json" -> generateJsonOutput outputDirectoryPath inputGraph - _ -> generateDotOutput outputDirectoryPath inputGraph + _ -> generateDotOutput outputDirectoryPath withRank inputGraph -generateDotOutput :: String -> Gr SimpleTaxon Double -> IO () -generateDotOutput outputDirectoryPath inputGraph = do - let diagram = drawTaxonomy (grev inputGraph) +generateDotOutput :: String -> Bool -> Gr SimpleTaxon Double -> IO () +generateDotOutput outputDirectoryPath withRank inputGraph = do + let diagram = drawTaxonomy withRank (grev inputGraph) writeFile (outputDirectoryPath ++ "taxonomy.dot") diagram @@ -86,12 +90,14 @@ -- | Extract taxids from RNAlien result.csv -extractTaxidsAlienCSV :: String -> IO [Node] +extractTaxidsAlienCSV :: String -> IO (Either String (V.Vector (Int,L.ByteString,L.ByteString))) extractTaxidsAlienCSV alienCSVPath = do let myOptions = defaultDecodeOptions { decDelimiter = fromIntegral (ord ';') } inputCSV <- L.readFile alienCSVPath - let decodedCsvOutput = V.toList (fromRight (decodeWith myOptions HasHeader inputCSV :: Either String (V.Vector (Int,L.ByteString,L.ByteString)))) - let taxnodes = map (\(a,_,_) -> a :: Node) decodedCsvOutput - return taxnodes + let decodedCsvOutput = decodeWith myOptions HasHeader inputCSV :: Either String (V.Vector (Int,L.ByteString,L.ByteString)) + return decodedCsvOutput + +firstOfTaxCSVTriple :: (Int, L.ByteString, L.ByteString) -> Node +firstOfTaxCSVTriple (a,_,_) = a diff --git a/src/Bio/TaxIds2TreeCompare.hs b/src/Bio/TaxIds2TreeCompare.hs --- a/src/Bio/TaxIds2TreeCompare.hs +++ b/src/Bio/TaxIds2TreeCompare.hs @@ -19,7 +19,8 @@ data Options = Options { taxDumpDirectoryPath :: String, taxNodeCSVFilePath :: String, - levels :: Int, + levels :: Int, + withRank :: Bool, outputDirectoryPath :: String } deriving (Show,Data,Typeable) @@ -28,6 +29,7 @@ { taxDumpDirectoryPath = def &= name "i" &= help "Path to input NCBI taxonomy dump files directory", taxNodeCSVFilePath = def &= name "t" &= help "Path to input taxonomy csv, each column with comma separated taxids represents one tree", levels = (1 ::Int) &= name "l" &= help "Number defining maximum distance from root for nodes in subtree.", + withRank = True &= name "w" &= help "Add taxonomic ranks to output. Default: True", outputDirectoryPath = def &= name "o" &= help "Path to output directory" } &= summary "TaxIds2TreeCompare - Multiple lists of taxonomy ids are converted into a visualisation of the taxonomic tree highlighting the input nodes corresponding to their list." &= help "Florian Eggenhofer - 2015" &= verbosity @@ -43,7 +45,7 @@ treesTaxids <- extractTreesTaxidsCSV taxNodeCSVFilePath let subgraphs = map (\treeTaxids -> extractTaxonomySubTreebyLevel treeTaxids graph (Just levels)) treesTaxids let comparisonGraph = compareSubTrees subgraphs - let treeComparison = drawTreeComparison comparisonGraph + let treeComparison = drawTaxonomyComparison withRank comparisonGraph writeFile (outputDirectoryPath ++ "comparison.dot") treeComparison -- | Extract taxids from RNAlien result.csv