diff --git a/README.org b/README.org new file mode 100644 --- /dev/null +++ b/README.org @@ -0,0 +1,33 @@ +* convert-annotation + +Very early stages. Mix between =HTTP= and =wreq= libraries, need to fully +convert to =wreq= in the future. + +Each mode, =info= or =annotation=, has its own help. + +Main usage to convert, say, some unknown annotation to Ensembl annotation in a +=csv=: + +#+BEGIN_SRC sh +cat input.csv | convert-annotation annotation --database "Ensembl" --column "gene" +#+END_SRC + +#+BEGIN_SRC sh :exports results :results value code +convert-annotation -h +#+END_SRC + +#+RESULTS: +#+BEGIN_SRC sh +convert-annotation, Gregory W. Schwartz. Converts an unknown annotation to some +other annotation. + +Usage: convert-annotation (info | annotation) + +Available options: + -h,--help Show this help text + +Available commands: + info + annotation +#+END_SRC + diff --git a/app/Main.hs b/app/Main.hs --- a/app/Main.hs +++ b/app/Main.hs @@ -4,6 +4,7 @@ Converts an unknown annotation to Ensembl's annotation, or other annotation. -} +{-# LANGUAGE BangPatterns #-} {-# LANGUAGE DataKinds #-} {-# LANGUAGE DeriveGeneric #-} {-# LANGUAGE OverloadedStrings #-} @@ -17,6 +18,7 @@ import Data.List import Control.Monad import GHC.Generics +import Data.Semigroup -- Cabal import qualified Data.Text as T @@ -28,30 +30,42 @@ import Pipes.Csv import Options.Generic +import qualified Foreign.R as R +import Foreign.R (SEXP, SEXPTYPE) +import Language.R.Instance as R +import Language.R.QQ +import Language.R.Literal as R + -- Local import Types import EnsemblConvert import HUGOConvert import UniProtConvert +import RGeneConvert +import MSigDBRDataConvert -- | Command line arguments data Options = Info { delimiter :: Maybe String "([,] | CHAR) The delimiter of the CSV file." , database :: String - "(Ensembl | HUGO TYPE | UniProt) Which database to convert with. TYPE is the type of the original gene symbol. The compatible list is in http://www.genenames.org/help/rest-web-service-help. HUGO is only supported for Annotation." - , descriptionField :: String + "(Ensembl | HUGO TYPE | UniProt | RGene (TYPE, TYPE) | MSigDBRdata (FILE, RDATA, TYPE)) Which database to convert with. TYPE is the type of the original gene symbol. The compatible list for TYPE with HUGO is in http://www.genenames.org/help/rest-web-service-help. HUGO is only supported for Annotation. RGene (Annotation only) takes in a type of (FROM, TO) for the gene symbol origin and destination. MSigDBRdata (Info only) takes an rdata file (tested with http://bioinf.wehi.edu.au/software/MSigDB/), the name of the rdata object containing the named list, and the TYPE of symbol (compatable list at http://bioconductor.org/packages/release/bioc/manuals/biomaRt/man/biomaRt.pdf in getGene) which returns pathways separated by \"/\"." + , descriptionField :: Maybe String "(Other TEXT | Description | Synonyms) The info to retrieve about the identifier. Description provides information about the identifier while synonyms provides alternate identifiers for the same entity. Returns a list of information (delimited by '/') for each match to Ensembl's cross references. For UniProt, enter a valid column (http://www.uniprot.org/help/programmatic_access)." , column :: T.Text "(COLUMN) The column containing the identifier. Must be a valid id for info." + , newColumn :: Maybe T.Text + "([Nothing] | COLUMN) The new column to put the results into. If unspecified, replaces the original column." , remove :: Bool "Whether to remove empty results (no matches to the database)." } | Annotation { delimiter :: Maybe String "([,] | CHAR) The delimiter of the CSV file." , database :: String - "(Ensembl | HUGO TYPE | UniProt) Which database to convert with. TYPE is the type of the original gene symbol. The compatible list is in http://www.genenames.org/help/rest-web-service-help. HUGO is only supported for Annotation." + "(Ensembl | HUGO TYPE | UniProt | RGene (TYPE, TYPE) | MSigDBRdata (FILE, RDATA, TYPE)) Which database to convert with. TYPE is the type of the original gene symbol. The compatible list for TYPE with HUGO is in http://www.genenames.org/help/rest-web-service-help. HUGO is only supported for Annotation. RGene (Annotation only) takes in a type of (FROM, TO) for the gene symbol origin and destination. MSigDBRdata (Info only) takes an rdata file (tested with http://bioinf.wehi.edu.au/software/MSigDB/), the name of the rdata object containing the named list, and the TYPE of symbol (compatable list at http://bioconductor.org/packages/release/bioc/manuals/biomaRt/man/biomaRt.pdf in getGene) which returns pathways separated by \"/\"." , column :: T.Text "(COLUMN) The column containing the identifier. Must be a valid id for info." + , newColumn :: Maybe T.Text + "([Nothing] | COLUMN) The new column to put the results into. If unspecified, replaces the original column." , remove :: Bool "Whether to remove empty results (no matches to the database)." } @@ -61,20 +75,25 @@ -- | Map the header column to the rest of the file for converting that -- column. -pipeConvert :: Options -> Pipe [T.Text] [T.Text] IO () -pipeConvert opts = do +pipeConvert :: Options + -> Maybe (RMart s) + -> Maybe (RData s) + -> Pipe [T.Text] [T.Text] IO () +pipeConvert opts rMart rData = do h <- await - let c = col opts h + let c = col opts h + newCol = unHelpful . newColumn $ opts - yield h + yield . maybe h (\x -> h <> [x]) $ newCol forever $ do x <- await - newX <- lift . convert opts . (!! c) $ x - unless - ((unHelpful . remove $ opts) && T.null newX) - (yield . L.set (L.ix c) newX $ x) + newX <- lift . convert opts rMart rData . (!! c) $ x + unless ((unHelpful . remove $ opts) && T.null newX) + . maybe (yield . L.set (L.ix c) newX $ x) + (const (yield (x <> [newX]))) + $ newCol return () return () @@ -85,23 +104,41 @@ fromMaybe (error "Column not found.") . elemIndex (unHelpful $ column opts) -- | The conversion process. -convert :: Options -> T.Text -> IO T.Text -convert opts@(Info { descriptionField = df }) = +convert :: Options -> Maybe (RMart s) -> Maybe (RData s) -> T.Text -> IO T.Text +convert opts@(Info { descriptionField = df }) rMart rData = fmap (fromMaybe "" . fmap unDesc) . whichDesc (read . unHelpful . database $ opts) . UnknownAnn where - whichDesc Ensembl = toEnsemblDesc (read . unHelpful $ df) + whichDesc Ensembl = toEnsemblDesc ( read + . fromMaybe (error "Needs description field.") + . unHelpful + $ df + ) whichDesc (HUGO _) = error "HUGO description not yet supported." - whichDesc UniProt = toUniProtDesc (read . unHelpful $ df) -convert opts@(Annotation {}) = + whichDesc UniProt = toUniProtDesc ( read + . fromMaybe (error "Needs description field.") + . unHelpful + $ df + ) + whichDesc (RGene _) = error "RGene description not yet supported." + whichDesc (MSigDBRData queryType) = + toMSigDBPathways + (fromJust rData) + (fromJust rMart) + (MSigDBType queryType) +convert opts@(Annotation {}) rMart rData = fmap (fromMaybe "" . fmap unAnn) . whichAnn (read . unHelpful . database $ opts) . UnknownAnn where - whichAnn Ensembl = toEnsemblAnn - whichAnn (HUGO queryType) = toHUGOAnn . HUGOType $ queryType - whichAnn UniProt = toUniProtAnn + whichAnn Ensembl = toEnsemblAnn + whichAnn (HUGO queryType) = toHUGOAnn . HUGOType $ queryType + whichAnn UniProt = toUniProtAnn + whichAnn (RGene queryType) = + toRGeneAnn (fromJust rMart) (RType queryType) + whichAnn (MSigDBRData _) = + error "MSigDBRData annotation not yet supported." main :: IO () main = do @@ -118,10 +155,20 @@ csvOpts = CSV.defaultDecodeOptions { CSV.decDelimiter = fromIntegral (ord delim) } - runEffect $ decodeWith csvOpts NoHeader PB.stdin + R.withEmbeddedR R.defaultConfig $ R.runRegion $ do + rMart <- case read . unHelpful . database $ opts of + (RGene _) -> fmap Just getRMart + (MSigDBRData _) -> fmap Just getRMart + _ -> return Nothing + rData <- case read . unHelpful . database $ opts of + (MSigDBRData (!file, !object, _)) -> + fmap Just . getRData (File file) $ object + _ -> return Nothing + + liftIO $ runEffect $ decodeWith csvOpts NoHeader PB.stdin >-> P.concat - >-> (pipeConvert opts) + >-> (pipeConvert opts rMart rData) >-> encode >-> PB.stdout - return () + return () diff --git a/convert-annotation.cabal b/convert-annotation.cabal --- a/convert-annotation.cabal +++ b/convert-annotation.cabal @@ -1,7 +1,7 @@ name: convert-annotation -version: 0.2.0.1 +version: 0.4.0.0 synopsis: Convert the annotation of a gene to another in a delimited file using a variety of different databases. -description: Please see README.md +description: Please see README.org homepage: http://github.com/GregorySchwartz/convert-annotation#readme license: GPL-3 license-file: LICENSE @@ -10,7 +10,7 @@ copyright: Copyright: (c) 2016 Gregory W. Schwartz category: Bioinformatics build-type: Simple --- extra-source-files: +extra-source-files: README.org cabal-version: >=1.10 library @@ -19,6 +19,8 @@ , EnsemblConvert , HUGOConvert , UniProtConvert + , RGeneConvert + , MSigDBRDataConvert build-depends: base >= 4.7 && < 5 , containers , bytestring @@ -29,6 +31,7 @@ , wreq , HTTP , safe + , inline-r ghc-options: -O2 default-language: Haskell2010 @@ -47,6 +50,7 @@ , pipes-bytestring , pipes-csv , lens + , inline-r default-language: Haskell2010 source-repository head diff --git a/src/MSigDBRDataConvert.hs b/src/MSigDBRDataConvert.hs new file mode 100644 --- /dev/null +++ b/src/MSigDBRDataConvert.hs @@ -0,0 +1,64 @@ +{- MSigDBRDataConvert +Gregory W. Schwartz + +Collections the functions pertaining to converting certain annotations into +pathways using the MSigDB rdata files (tested with +http://bioinf.wehi.edu.au/software/MSigDB/). +-} + +{-# LANGUAGE BangPatterns #-} +{-# LANGUAGE OverloadedStrings #-} +{-# LANGUAGE QuasiQuotes #-} + +module MSigDBRDataConvert + ( getRData + , toMSigDBPathways + ) where + +-- Standard + +-- Cabal +import qualified Data.Text as T + +import qualified Foreign.R as R +import Foreign.R (SEXP, SEXPTYPE) +import Language.R.Instance as R +import Language.R.QQ +import Language.R.Literal as R +import H.Prelude + +-- Local +import Types +import RGeneConvert + +-- | Get the RData object. +getRData :: File -> String -> R s (RData s) +getRData (File file) object = fmap RData + $ [r| load(file_hs) + res = get(object_hs) + res + |] + +-- | Get the R mapping of gene to gene. +toMSigDBPathways + :: RData s + -> RMart s + -> MSigDBType + -> UnknownAnn + -> IO (Maybe Desc) +toMSigDBPathways _ _ _ (UnknownAnn "") = return Nothing +toMSigDBPathways rData rMart (MSigDBType (_, _, !from)) query = + (fmap . fmap) Desc $ R.runRegion $ do + entrez <- io . toRGeneAnn rMart (RType (from, "entrezgene")) $ query + let object = unRData rData + + case entrez of + Nothing -> return Nothing + (Just (Ann gText)) -> do + let g = T.unpack gText + res <- [r| pathNames = names(object_hs[unlist(lapply(object_hs, function(x) (g_hs %in% unlist(x))))]) |] + let pathNames = R.fromSomeSEXP res :: [String] + if null . drop 1 $ pathNames + then return Nothing + else + return . Just . T.intercalate "/" . fmap T.pack $ pathNames diff --git a/src/RGeneConvert.hs b/src/RGeneConvert.hs new file mode 100644 --- /dev/null +++ b/src/RGeneConvert.hs @@ -0,0 +1,57 @@ +{- RGeneConvert +Gregory W. Schwartz + +Collections the functions pertaining to converting certain annotations into +other annotations using biomart from R +(http://bioconductor.org/packages/release/bioc/html/biomaRt.html). +-} + +{-# LANGUAGE BangPatterns #-} +{-# LANGUAGE OverloadedStrings #-} +{-# LANGUAGE QuasiQuotes #-} + +module RGeneConvert + ( getRMart + , toRGeneAnn + ) where + +-- Standard + +-- Cabal +import qualified Data.Text as T + +import qualified Foreign.R as R +import Foreign.R (SEXP, SEXPTYPE) +import Language.R.Instance as R +import Language.R.QQ +import Language.R.Literal as R + +-- Local +import Types + +-- | Get the R mapping of gene to gene. +getRMart :: R s (RMart s) +getRMart = fmap RMart + $ [r| library(biomaRt) + mart = useMart("ensembl", dataset="hsapiens_gene_ensembl") + |] + +-- | Get the R mapping of gene to gene. +toRGeneAnn :: RMart s -> RType -> UnknownAnn -> IO (Maybe Ann) +toRGeneAnn _ _ (UnknownAnn "") = return Nothing +toRGeneAnn rMart (RType (!from, !to)) (UnknownAnn textQuery) = + (fmap . fmap) Ann $ R.runRegion $ do + let query = T.unpack textQuery + mart = unRMart rMart + res <- [r| map = getBM( attributes = c(from_hs, to_hs) + , filters = from_hs + , values = c(query_hs) + , mart = mart_hs + , uniqueRows=FALSE + ) + as.character(map[1,2]) + |] + + let naCheck "NA" = Nothing + naCheck x = Just x + return . fmap T.pack . naCheck $ (R.fromSomeSEXP res :: String) diff --git a/src/Types.hs b/src/Types.hs --- a/src/Types.hs +++ b/src/Types.hs @@ -15,20 +15,36 @@ import qualified Data.Text as T import Data.Aeson +import qualified Foreign.R as R +import Foreign.R (SEXP, SEXPTYPE) +import Language.R.Instance as R +import Language.R.QQ + -- Local -- Algebraic -data Database = Ensembl | HUGO T.Text | UniProt deriving (Read, Show) +data Database + = Ensembl + | HUGO T.Text + | UniProt + | RGene (String, String) + | MSigDBRData (String, String, String) + deriving (Read,Show) data DescFields = UniProtOther T.Text | Synonyms | Description deriving (Read,Show) -- Basic +newtype File = File String newtype UnknownAnn = UnknownAnn { unUnknownAnn :: T.Text } newtype Ann = Ann { unAnn :: T.Text } newtype Desc = Desc { unDesc :: T.Text } newtype HUGOType = HUGOType { unHUGOType :: T.Text } +newtype RType = RType { unRType :: (String, String) } +newtype MSigDBType = MSigDBType { unMSigDBType :: (String, String, String) } +newtype RMart s = RMart { unRMart :: (R.SomeSEXP s) } +newtype RData s = RData { unRData :: (R.SomeSEXP s) } -- Advanced