diff --git a/RNAlien.cabal b/RNAlien.cabal --- a/RNAlien.cabal +++ b/RNAlien.cabal @@ -1,8 +1,5 @@ --- Initial RNAlien.cabal generated by cabal init. For further documentation, --- see http://haskell.org/cabal/users-guide/ - name: RNAlien -version: 1.2.6 +version: 1.2.8 synopsis: Unsupervized construction of RNA family models description: RNAlien is a tool for automatic construction of RNAfamily models from a single sequence. . @@ -53,8 +50,8 @@ source-repository this type: git - location: https://github.com/eggzilla/RNAlien/tree/1.2.6 - tag: 1.2.6 + location: https://github.com/eggzilla/RNAlien/tree/1.2.8 + tag: 1.2.8 executable RNAlien Hs-Source-Dirs: ./src/Bio/ @@ -89,7 +86,7 @@ Library Hs-Source-Dirs: ./src/ ghc-options: -Wall -O2 -fno-warn-unused-do-bind - build-depends: base >=4.5 && <5, cmdargs, ViennaRNAParser>=1.3.1, process, directory, blastxml>=0.3.2, biofasta, parsec, random, BlastHTTP>=1.2.1, biocore, bytestring, Taxonomy >= 1.0.2, either-unwrap, containers, ClustalParser>=1.1.0, EntrezHTTP>=1.0.3, vector, edit-distance, cassava, matrix, hierarchical-clustering, filepath, HTTP, http-conduit, hxt, network, aeson, text, transformers, pureMD5 + build-depends: base >=4.5 && <5, cmdargs, ViennaRNAParser>=1.3.1, process, directory, blastxml>=0.3.2, biofasta, parsec, random, BlastHTTP>=1.2.1, biocore, bytestring, Taxonomy >= 1.0.2, either-unwrap, containers, ClustalParser>=1.1.0, EntrezHTTP>=1.0.3, vector, edit-distance, cassava, matrix, hierarchical-clustering, filepath, HTTP, http-conduit, hxt, network, aeson, text, transformers, pureMD5, http-types Exposed-Modules: Bio.RNAlienData Bio.RNAlienLibrary Bio.RNAcentralHTTP diff --git a/src/Bio/RNAcentralHTTP.hs b/src/Bio/RNAcentralHTTP.hs --- a/src/Bio/RNAcentralHTTP.hs +++ b/src/Bio/RNAcentralHTTP.hs @@ -9,8 +9,8 @@ buildStringViaMD5Query, getRNACentralEntries, showRNAcentralAlienEvaluation, - RNAcentralEntryResponse, - RNAcentralEntry + RNAcentralEntryResponse(..), + RNAcentralEntry(..) ) where import Network.HTTP.Conduit @@ -24,7 +24,6 @@ import Bio.Core.Sequence import Bio.Sequence.Fasta import Data.Either -import Data.Aeson.Types --Datatypes -- | Data structure for RNAcentral entry response diff --git a/src/Bio/RNAcentralHTTPRequest.hs b/src/Bio/RNAcentralHTTPRequest.hs --- a/src/Bio/RNAcentralHTTPRequest.hs +++ b/src/Bio/RNAcentralHTTPRequest.hs @@ -6,8 +6,6 @@ module Main where import System.Console.CmdArgs -import Bio.RNAlienLibrary -import Data.Either.Unwrap import Bio.RNAcentralHTTP data Options = Options diff --git a/src/Bio/RNAlien.hs b/src/Bio/RNAlien.hs --- a/src/Bio/RNAlien.hs +++ b/src/Bio/RNAlien.hs @@ -33,7 +33,8 @@ inputQueryNumber :: Int, threads :: Int, taxonomyRestriction :: Maybe String, - sessionIdentificator :: Maybe String + sessionIdentificator :: Maybe String, + performEvaluation :: Bool } deriving (Show,Data,Typeable) options :: Options @@ -52,8 +53,9 @@ inputQueryNumber = (5 :: Int) &= name "n" &= help "Number of queries used for candidate search. Default: 5", threads = 1 &= name "c" &= help "Number of available cpu slots/cores. Default: 1", taxonomyRestriction = Nothing &= name "r" &= help "Restrict search space to taxonomic kingdom (bacteria,archea,eukaryia). Default: not set", - sessionIdentificator = Nothing &= name "d" &= help "Optional session id that is used instead of automatically generated one." - } &= summary ("RNAlien " ++ alienVersion) &= help "Florian Eggenhofer, Ivo L. Hofacker, Christian Höner zu Siederdissen - 2013 - 2016" &= verbosity + sessionIdentificator = Nothing &= name "d" &= help "Optional session id that is used instead of automatically generated one.", + performEvaluation = True &= name "x" &= help "Perform evaluation step. Default: True" + } &= summary ("RNAlien " ++ alienVersion) &= help "Florian Eggenhofer, Ivo L. Hofacker, Christian Höner zu Siederdissen - 2013 - 2017" &= verbosity main :: IO () main = do @@ -100,11 +102,16 @@ logMessage (show initialization) temporaryDirectoryPath modelConstructionResults <- modelConstructer staticOptions initialization let resultTaxonomyRecordsCSVTable = constructTaxonomyRecordsCSVTable modelConstructionResults - resultEvaluation <- evaluateConstructionResult staticOptions modelConstructionResults - appendFile (temporaryDirectoryPath ++ "Log") resultEvaluation writeFile (temporaryDirectoryPath ++ "result.csv") resultTaxonomyRecordsCSVTable - resultSummary modelConstructionResults staticOptions - writeFile (temporaryDirectoryPath ++ "done") "" + if performEvaluation + then do + resultEvaluation <- evaluateConstructionResult staticOptions modelConstructionResults + appendFile (temporaryDirectoryPath ++ "Log") resultEvaluation + resultSummary modelConstructionResults staticOptions + writeFile (temporaryDirectoryPath ++ "done") "" + else do + resultSummary modelConstructionResults staticOptions + writeFile (temporaryDirectoryPath ++ "done") "" alienVersion :: String alienVersion = showVersion version diff --git a/src/Bio/RNAlienLibrary.hs b/src/Bio/RNAlienLibrary.hs --- a/src/Bio/RNAlienLibrary.hs +++ b/src/Bio/RNAlienLibrary.hs @@ -43,6 +43,7 @@ import Data.Int (Int16) import Bio.RNAlienData import qualified Data.ByteString.Lazy.Char8 as L +import qualified Data.ByteString.Char8 as B import Bio.Taxonomy import Data.Either.Unwrap import Data.Maybe @@ -63,9 +64,10 @@ import Bio.RNAfoldParser import Bio.RNAalifoldParser import Bio.RNAzParser -import Network.HTTP +import qualified Network.HTTP.Conduit as N +import Network.HTTP.Types.Status import qualified Bio.RNAcodeParser as RC -import Bio.RNAcentralHTTP +import qualified Bio.RNAcentralHTTP as RCH import Bio.InfernalParser import qualified Data.Text as T import qualified Data.Text.IO as TI @@ -1509,9 +1511,9 @@ let cmFilepath = tempDirPath staticOptions ++ "result.cm" let resultSequences = (inputFasta mCResult):map nucleotideSequence (concatMap sequenceRecords (taxRecords mCResult)) let resultNumber = length resultSequences + 1 - let rnaCentralQueries = map buildSequenceViaMD5Query resultSequences - rnaCentralEntries <- getRNACentralEntries rnaCentralQueries - let rnaCentralEvaluationResult = showRNAcentralAlienEvaluation rnaCentralEntries + let rnaCentralQueries = map RCH.buildSequenceViaMD5Query resultSequences + rnaCentralEntries <- RCH.getRNACentralEntries rnaCentralQueries + let rnaCentralEvaluationResult = RCH.showRNAcentralAlienEvaluation rnaCentralEntries writeFile (tempDirPath staticOptions ++ "result.rnacentral") rnaCentralEvaluationResult systemCMalign ("--outformat=Clustal --cpu " ++ show (cpuThreads staticOptions)) cmFilepath fastaFilepath clustalFilepath let resultModelStatistics = tempDirPath staticOptions ++ "result.cmstat" @@ -1652,14 +1654,13 @@ -- | Check if alien can connect to NCBI checkNCBIConnection :: IO (Either String String) checkNCBIConnection = do - response <- simpleHTTP (getRequest "https://www.ncbi.nlm.nih.gov") - if isRight response - then do - let rightResponse = fromRight response - if rspCode rightResponse == (2,0,0) - then return (Right ("Network connection with NCBI server is ok: " ++ show (rspCode rightResponse))) - else return (Left ("Could not connect to NCBI server \"https://www.ncbi.nlm.nih.gov\". Response Code: " ++ show (rspCode rightResponse))) - else return (Left ("Could not connect to NCBI server: \"https://www.ncbi.nlm.nih.gov\": " ++ show (fromLeft response))) + req <- N.parseRequest "https://www.ncbi.nlm.nih.gov" + manager <- N.newManager N.tlsManagerSettings + response <- N.httpLbs req manager + let sta = N.responseStatus response + if statusIsSuccessful sta + then return (Right ("Network connection with NCBI server was successful")) + else return (Left ("Could not connect to NCBI server \"https://www.ncbi.nlm.nih.gov\". Response Code: " ++ show (statusCode sta) ++ " \n" ++ B.unpack (statusMessage sta))) -- | Blast evalue is set stricter in inital alignment mode setBlastExpectThreshold :: ModelConstruction -> Double