biopsl 0.2 → 0.3

raw patch · 5 files changed

+110/−4 lines, 5 filesdep +cmdargsnew-component:exe:pslcovnew-component:exe:pslfilterPVP ok

version bump matches the API change (PVP)

Dependencies added: cmdargs

API changes (from Hackage documentation)

Files

• biopsl.cabal +18/−2
• examples/PslCov.hs +23/−0
• examples/PslFilter.hs +61/−0
• examples/PslUniq.hs +1/−2
• src/Bio/Alignment/PSL.hs +7/−0

@@ -1,5 +1,5 @@ Name:                biopsl-Version:             0.2+Version:             0.3 Synopsis:            Library and executables for working with PSL files Description:         The library contains the functionality for reading and writing 		     PSL files (alignment data, e.g. from BLAT output). It duplicates@@ -11,8 +11,13 @@ Stability:           Experimental Category:            Bioinformatics Build-type:          Simple-Cabal-version:       >=1.2+Cabal-version:       >=1.6 +Source-repository this+   Type:      darcs+   Location:  http://malde.org/~ketil/biohaskell/biosff+   Tag:       0.3+ Library   Exposed-modules: Bio.Alignment.PSL   Build-depends:   base >= 3 && < 5, biocore >= 0.1, bytestring@@ -21,5 +26,16 @@  Executable psluniq   Main-Is:        PslUniq.hs+  Hs-Source-Dirs: src, examples+  Ghc-Options:    -Wall++Executable pslfilter+  Main-Is:        PslFilter.hs+  Hs-Source-Dirs: src, examples+  Build-Depends:  cmdargs+  Ghc-Options:    -Wall++Executable pslcov+  Main-Is:        PslCov.hs   Hs-Source-Dirs: src, examples   Ghc-Options:    -Wall

@@ -0,0 +1,23 @@+-- PslCov: extract complete hits (more than x% of query covered)++module Main where++import Bio.Alignment.PSL+import System.Environment (getArgs)++main :: IO ()+main = do+  args <- getArgs+  case args of +    [pq,pt,f] -> printPSL . filter (isHit (read pq) (read pt)) =<< readPSL f+    _ -> error ("Usage: pslcov p q pslfile\n"+++                "where p is the minimum query coverage\n"+++                "and q is the minium target coverage.")++isHit :: Double -> Double -> PSL -> Bool+isHit p q x = m / qs > p && m / ts > q+  where qs = fromIntegral $ qsize x+        ts = fromIntegral $ tsize x+        m  = fromIntegral $ match x+         +

@@ -0,0 +1,61 @@+{-| Select a subset of PSL records according to specified criteria:+  e.g. pslfilter --identity-min --overhang-max ... --count-max+-}++{-# Language DeriveDataTypeable, OverloadedStrings #-}++import Bio.Alignment.PSL+import System.Console.CmdArgs+import qualified Data.ByteString.Lazy as L++data Options = Opts +  { input :: Maybe FilePath+  , identity, queryCov, targetCov :: Maybe Double+  , overhang :: Maybe Int+  } deriving (Data,Typeable)++defs :: Options+defs = Opts +  { input = Nothing &= args+  , identity = Nothing &= help "minimum identiy for matches" &= name "i"+  , overhang = Nothing &= help "maximum overhang for matches" &= name "h"+  , queryCov = Nothing &= help "minimum query coverage" &= name "q"+  , targetCov = Nothing &= help "minimum target coverage" &= name "t"+  } &= summary "pslfilter - extract a subset of PSL records"+    &= program "pslfilter"++main :: IO ()+main = do+  opts <- cmdArgs defs+  ps <- case input opts of +        Nothing  -> (parsePSL `fmap` L.getContents)+        Just f -> readPSL f+  let idfilter = case identity opts of+        Nothing -> const True+        Just d  -> (\x -> local_identity x >= d)+      ohfilter = case overhang opts of  +        Nothing -> const True+        Just o  -> (\x -> (uncurry max $ overhangs x) < o)+      qcfilter = case queryCov opts of+          Nothing -> const True+          Just q -> (\x -> fromIntegral (match x) / fromIntegral (qsize x) >= q)+      tcfilter = case targetCov opts of+          Nothing -> const True+          Just t -> (\x -> fromIntegral (match x) / fromIntegral (tsize x) >= t)+  printPSL $ filter (idfilter .&. ohfilter .&. qcfilter .&. tcfilter ) ps+++(.&.) :: (t -> Bool) -> (t -> Bool) -> t -> Bool+(.&.) f g = \x -> f x && g x++-- | Calculate the smallest overhang, i.e. the unmatched parts+--   of either query or target on each side of the local alignment.+overhangs :: PSL -> (Int,Int)+overhangs p = case strand p of+  "+" -> (min (qstart p) (tstart p), min (qsize p-qend p) (tsize p-tend p))+  "-" -> (min (qstart p) (tsize p-tend p), min (qsize p-qend p) (tstart p))+  _ -> error ("I couldn't understand strand :"++show (strand p))++-- | Calculate the identity score of the matched region+local_identity :: PSL -> Double+local_identity p = fromIntegral (match p) / fromIntegral (match p + mismatch p)

@@ -13,8 +13,7 @@         [] -> (parsePSL `fmap` L.getContents)         [f] -> readPSL f         _ -> error "Usage: psluniq [pslfile]"-  L.putStr pslHeader-  L.putStr $ unparsePSL $ uniq ps+  printPSL $ uniq ps    uniq :: [PSL] -> [PSL] uniq [] = []

@@ -4,6 +4,8 @@      See <http://genome.ucsc.edu/FAQ/FAQformat#format2> for details. -}++{-# Options -funbox-strict-fields #-} -- makes no difference module Bio.Alignment.PSL where  import qualified Data.ByteString.Lazy.Char8 as B@@ -31,6 +33,11 @@ -- | Create a PSL file from a list of alignments. writePSL :: FilePath -> [PSL] -> IO () writePSL f =  B.writeFile f . B.append pslHeader . unparsePSL ++printPSL :: [PSL] -> IO ()+printPSL ps = do+  B.putStr pslHeader+  B.putStr $ unparsePSL ps  -- | Parse a 'ByteString' as a PSL file (note that it must contain the PSL header). parsePSL :: ByteString -> [PSL]