diff --git a/Biobase/Turner.hs b/Biobase/Turner.hs --- a/Biobase/Turner.hs +++ b/Biobase/Turner.hs @@ -1,3 +1,4 @@ +{-# LANGUAGE TypeOperators #-} -- | The 'Turner2004' data structure reflects the RNA (and DNA) energy -- parameters known as the Turner 2004 data set. @@ -10,41 +11,48 @@ import Data.ByteString import Data.Map as M +import Data.Array.Repa.Index import Biobase.Primary import Biobase.Secondary import Data.PrimitiveArray -import Data.PrimitiveArray.Ix +import Data.PrimitiveArray.Unboxed -- | The parameters. Turner parameters are set by the Import module for -- nucleotides n,a,c,g,u. All values that are not read (or are ".") will end up -- with a value > 100K. -- --- [1] Yes, such instances are easily created, but I don't want to pull in --- another library and I don't want to create an Ix instance here. --- -- TODO use 'Energy' instead of 'Double' +-- +-- TODO specialized shape types for pairs? +type PP = (Z:.Nuc:.Nuc:.Nuc:.Nuc) +type PN = (Z:.Nuc:.Nuc:.Nuc) +type PNN = (Z:.Nuc:.Nuc:.Nuc:.Nuc) +type PPNN = PP:.Nuc:.Nuc +type PPNNN = PPNN:.Nuc +type PPNNNN = PPNNN:.Nuc + data Turner2004 = Turner2004 - { stack :: PrimArray (Pair,Pair) Double + { stack :: PrimArray PP Double , dangle3 :: PrimArray PN Double , dangle5 :: PrimArray PN Double - , hairpinL :: PrimArray Int Double + , hairpinL :: PrimArray DIM1 Double , hairpinMM :: PrimArray PNN Double , hairpinLookup :: M.Map ByteString Double , hairpinGGG :: Double , hairpinCslope :: Double , hairpinCintercept :: Double , hairpinC3 :: Double - , bulgeL :: PrimArray Int Double + , bulgeL :: PrimArray DIM1 Double , bulgeSingleC :: Double - , iloop1x1 :: PrimArray (Pair,Pair,(Nuc,Nuc)) Double - , iloop2x1 :: PrimArray (Pair,Pair,(Nuc,Nuc,Nuc)) Double - , iloop2x2 :: PrimArray (Pair,Pair,(Nuc,Nuc,Nuc,Nuc)) Double -- yeah, 6-tuple ix instances :-( [1] + , iloop1x1 :: PrimArray PPNN Double + , iloop2x1 :: PrimArray PPNNN Double + , iloop2x2 :: PrimArray PPNNNN Double , iloopMM :: PrimArray PNN Double , iloop2x3MM :: PrimArray PNN Double , iloop1xnMM :: PrimArray PNN Double - , iloopL :: PrimArray Int Double + , iloopL :: PrimArray DIM1 Double , multiMM :: PrimArray PNN Double , ninio :: Double , maxNinio :: Double @@ -54,13 +62,11 @@ , multiAsym :: Double , multiStrain :: Double , extMM :: PrimArray PNN Double - , coaxial :: PrimArray (Pair,Pair) Double -- no intervening unpaired nucleotides + , coaxial :: PrimArray PP Double -- no intervening unpaired nucleotides , coaxStack :: PrimArray PNN Double , tStackCoax :: PrimArray PNN Double , largeLoop :: Double , termAU :: Double , intermolecularInit :: Double - } deriving (Read,Show) + } deriving (Show) -type PNN = (Pair,Nuc,Nuc) -type PN = (Pair,Nuc) diff --git a/Biobase/Turner/Import.hs b/Biobase/Turner/Import.hs --- a/Biobase/Turner/Import.hs +++ b/Biobase/Turner/Import.hs @@ -1,3 +1,5 @@ +{-# LANGUAGE PatternGuards #-} +{-# LANGUAGE TupleSections #-} {-# LANGUAGE NoMonomorphismRestriction #-} {-# LANGUAGE OverloadedStrings #-} @@ -46,6 +48,7 @@ import qualified Data.List as L import System.FilePath.Posix import Data.Maybe (fromJust) +import Data.Array.Repa.Index import Biobase.Primary import Biobase.Secondary @@ -83,17 +86,17 @@ cstack' <- blockFromFile $ fp prefix ++ "coaxstack" <.> suffix tstack' <- blockFromFile $ fp prefix ++ "tstackcoax" <.> suffix return Turner2004 - { stack = fromAssocs minPP maxPP infE $ L.zip keysPP stack' - , dangle3 = fromAssocs minPB maxPB infE $ L.zip keysPB dangle3' - , dangle5 = fromAssocs minPB maxPB infE $ L.zip keysPB dangle5' - , hairpinL = fromAssocs 0 30 infE $ L.zip [1..30] hairpinL' + { stack = fromAssocs minPP maxPP infE $ L.zip keysPP stack' + , dangle3 = fromAssocs minPB maxPB infE $ L.zip keysPB dangle3' + , dangle5 = fromAssocs minPB maxPB infE $ L.zip keysPB dangle5' + , hairpinL = fromAssocs (Z:.0) (Z:.30) infE $ L.zip d1_30 hairpinL' , hairpinMM = fromAssocs minPBB maxPBB infE $ L.zip keysPBB hairpinMM' , hairpinLookup = M.fromList $ hairpinLk3 ++ hairpinLk4 ++ hairpinLk6 , hairpinGGG = L.head $ imisc' !! 8 , hairpinCslope = L.head $ imisc' !! 9 , hairpinCintercept = L.head $ imisc' !! 10 , hairpinC3 = L.head $ imisc' !! 11 - , bulgeL = fromAssocs 0 30 infE $ L.zip [1..30] bulgeL' + , bulgeL = fromAssocs (Z:.0) (Z:.30) infE $ L.zip d1_30 bulgeL' , bulgeSingleC = L.head $ imisc' !! 13 , iloop1x1 = fromAssocs minPPBB maxPPBB infE $ L.zip keysPPBB iloop1x1' , iloop2x1 = fromAssocs minPPBBB maxPPBBB infE $ L.zip keysPPBBB iloop2x1' @@ -101,7 +104,7 @@ , iloopMM = fromAssocs minPBB maxPBB infE $ L.zip keysPBB iloopMM' , iloop2x3MM = fromAssocs minPBB maxPBB infE $ L.zip keysPBB iloop2x3MM' , iloop1xnMM = fromAssocs minPBB maxPBB infE $ L.zip keysPBB iloop1xnMM' - , iloopL = fromAssocs 0 30 infE $ L.zip [1..30] iloopL' + , iloopL = fromAssocs (Z:.0) (Z:.30) infE $ L.zip d1_30 iloopL' , multiMM = fromAssocs minPBB maxPBB infE $ L.zip keysPBB multiMM' , ninio = L.head $ imisc' !! 2 , maxNinio = L.head $ imisc' !! 1 @@ -119,32 +122,34 @@ , intermolecularInit = L.head $ imisc' !! 12 } -minPP = (minP,minP) -maxPP = (maxP,maxP) -minP = (nN,nN) -maxP = (nU,nU) -minPB = (minP,nN) -maxPB = (maxP,nU) -minPBB = (minP,nN,nN) -maxPBB = (maxP,nU,nU) -minPPBB = (minP,minP,(nN,nN)) -maxPPBB = (maxP,maxP,(nU,nU)) -minPPBBB = (minP,minP,(nN,nN,nN)) -maxPPBBB = (maxP,maxP,(nU,nU,nU)) -minPPBBBB = (minP,minP,(nN,nN,nN,nN)) -maxPPBBBB = (maxP,maxP,(nU,nU,nU,nU)) +minPP = Z:.nN:.nN:.nN:.nN -- (minP,minP) +maxPP = Z:.nU:.nU:.nU:.nU -- (maxP,maxP) +minP = Z:.nN:.nN -- (nN,nN) +maxP = Z:.nU:.nU -- (nU,nU) +minPB = minP:.nN -- (minP,nN) +maxPB = maxP:.nU -- (maxP,nU) +minPBB = minPB:.nN -- (minP,nN,nN) +maxPBB = maxPB:.nU -- (maxP,nU,nU) +minPPBB = minPP:.nN:.nN -- (minP,minP,(nN,nN)) +maxPPBB = maxPP:.nU:.nU -- (maxP,maxP,(nU,nU)) +minPPBBB = minPPBB:.nN -- (minP,minP,(nN,nN,nN)) +maxPPBBB = maxPPBB:.nU -- (maxP,maxP,(nU,nU,nU)) +minPPBBBB = minPPBBB:.nN -- (minP,minP,(nN,nN,nN,nN)) +maxPPBBBB = maxPPBBB:.nU -- (maxP,maxP,(nU,nU,nU,nU)) -keysPP = [((k1,k2),(k4,k3)) | k1 <- acgu, k3 <- acgu, k2 <- acgu, k4 <- acgu] -keysPB = [((k1,k2),k3) | k1 <- acgu, k2 <- acgu, k3 <- acgu] -keysPBB = [ ((k1,k2),k3,k4) +d1_30 = L.map (Z:.) [1..30] + +keysPP = [{- ((k1,k2),(k4,k3)) -} Z:.k1:.k2:.k4:.k3 | k1 <- acgu, k3 <- acgu, k2 <- acgu, k4 <- acgu] +keysPB = [{- ((k1,k2),k3) -} Z:.k1:.k2:.k3 | k1 <- acgu, k2 <- acgu, k3 <- acgu] +keysPBB = [ {- ((k1,k2),k3,k4) -} Z:.k1:.k2:.k3:.k4 | k1 <- acgu, k3 <- acgu, k2 <- acgu, k4 <- acgu] -keysPPBB = [ ((k1,k2),(k4,k3),(k5,k6)) +keysPPBB = [ {- ((k1,k2),(k4,k3),(k5,k6)) -} Z:.k1:.k2:.k4:.k3:.k5:.k6 | (k1,k2) <- plist11, k5 <- acgu, (k3,k4) <- plist11, k6 <- acgu] -keysPPBBB = [ ((k1,k2),(k4,k3),(k5,k6,k7)) +keysPPBBB = [ {- ((k1,k2),(k4,k3),(k5,k6,k7)) -} Z:.k1:.k2:.k4:.k3:.k5:.k6:.k7 | (k1,k2) <- plist11, k6 <- acgu, k5 <- acgu, (k3,k4) <- plist11, k7 <- acgu] -keysPPBBBBrna = [ ((k1,k2),(k4,k3),(k5,k6,k7,k8)) +keysPPBBBBrna = [ {- ((k1,k2),(k4,k3),(k5,k6,k7,k8)) -} Z:.k1:.k2:.k4:.k3:.k5:.k6:.k7:.k8 | (k1,k2) <- plist22rna, (k3,k4) <- plist22rna, k5 <- acgu, k8 <- acgu, k6 <- acgu, k7 <- acgu] -keysPPBBBBdna = [ ((k1,k2),(k4,k3),(k5,k6,k7,k8)) +keysPPBBBBdna = [ {- ((k1,k2),(k4,k3),(k5,k6,k7,k8)) -} Z:.k1:.k2:.k4:.k3:.k5:.k6:.k7:.k8 | (k1,k2) <- plist22dna, (k3,k4) <- plist22dna, k5 <- acgu, k8 <- acgu, k6 <- acgu, k7 <- acgu] plist11 = [(nA,nU),(nC,nG),(nG,nC),(nU,nA),(nG,nU),(nU,nG)] diff --git a/BiobaseTurner.cabal b/BiobaseTurner.cabal --- a/BiobaseTurner.cabal +++ b/BiobaseTurner.cabal @@ -1,9 +1,9 @@ name: BiobaseTurner -version: 0.1.1.0 +version: 0.2.0.0 author: Christian Hoener zu Siederdissen maintainer: choener@tbi.univie.ac.at homepage: http://www.tbi.univie.ac.at/~choener/ -copyright: Christian Hoener zu Siederdissen, 2010,2011 +copyright: Christian Hoener zu Siederdissen, 2010-2012 category: Bioinformatics synopsis: Import Turner RNA parameters license: GPL-3 @@ -26,6 +26,10 @@ enumerations, which we handle rather... simplistically. We cannot fix one asymmetry case in dnastack.dh, as we do not change sources. + . + BIG FAT WARNING indexing now depends on repa shapes and index + representations. (with a good reason coming soon). For now, + just assume that this has performance benefits. library build-depends: @@ -36,13 +40,19 @@ filepath, iteratee, split, - BiobaseXNA, - PrimitiveArray + BiobaseXNA >= 0.6.0.0, + vector >= 0.9, + repa >= 2.0, + PrimitiveArray >= 0.1.1.2 exposed-modules: Biobase.Turner Biobase.Turner.Import ghc-options: - -O2 + -Odph -funbox-strict-fields -fspec-constr + +source-repository head + type: git + location: git://github.com/choener/BiobaseTurner