biopsl 0.3 → 0.4

raw patch · 3 files changed

+49/−2 lines, 3 filesdep +unordered-containersnew-component:exe:pslstatsPVP ok

version bump matches the API change (PVP)

Dependencies added: unordered-containers

API changes (from Hackage documentation)

Files

• biopsl.cabal +7/−1
• examples/PslStats.hs +34/−0
• examples/PslUniq.hs +8/−1

@@ -1,5 +1,5 @@ Name:                biopsl-Version:             0.3+Version:             0.4 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@@ -23,6 +23,12 @@   Build-depends:   base >= 3 && < 5, biocore >= 0.1, bytestring   Hs-Source-Dirs:  src   Ghc-Options:     -Wall++Executable pslstats+  Main-Is:        PslStats.hs+  Hs-Source-Dirs: src, examples+  Build-depends:  unordered-containers+  Ghc-Options:    -Wall  Executable psluniq   Main-Is:        PslUniq.hs

@@ -0,0 +1,34 @@+{-| Generate stats from PSL files, similar to 'samtools stats' for+    BAM files.  Each target sequence is listed with its length and +    number of matching query sequences.+-}++import Bio.Alignment.PSL+import System.Environment (getArgs)+import Data.HashMap.Strict as H hiding (map)+import qualified Data.ByteString.Lazy.Char8 as L+import Data.List (foldl', intercalate)++main :: IO ()+main = do+  fs <- getArgs+  ps <- mapM readPSL fs+  putStrLn ("target\tlenght\t"++intercalate "\t" fs)+  putStr $ unlines $ format $ countTargets ps++-- | map of lengths, and map of count per input file+countTargets :: [[PSL]] -> [HashMap L.ByteString Int]+countTargets = go [] empty+  where go cs ls (ps:pss) = let (l,c) = count1 ls ps+                            in go (c:cs) l pss+        go cs ls [] = (ls:reverse cs)+        count1 ls ps = Data.List.foldl' ins1 (ls,empty) ps+        ins1 (l,c) p = l `seq` c `seq` (H.insert (tname p) (tsize p) l,H.insertWith (+) (tname p) 1 c)+++format :: [HashMap L.ByteString Int] -> [String]+format (l:cs) = [unwords' $ (L.unpack s:show len:map (show . H.lookupDefault 0 s) cs) | (s,len) <- toList l]+format _ = error "format"++unwords' :: [String] -> String+unwords' = intercalate "\t"

@@ -5,6 +5,7 @@ import Bio.Alignment.PSL import System.Environment (getArgs) import qualified Data.ByteString.Lazy as L+import Data.List (sortBy)  main :: IO () main = do@@ -17,10 +18,16 @@    uniq :: [PSL] -> [PSL] uniq [] = []-uniq (p:ps) = go p ps+uniq (p:ps) +  -- protein matches are sorted first by strand, thus we need to resort to sorting+  | L.length (strand p) > 1 = let (q:qs) = sortOn qname (p:ps) in go q qs+  | otherwise               = go p ps    go :: PSL -> [PSL] -> [PSL] go p1 (q:qs) | qname q /= qname p1 = p1 : go q qs              | match q > match p1  = go q qs              | otherwise           = go p1 qs go p1 [] = [p1]++sortOn :: Ord a1 => (a -> a1) -> [a] -> [a]+sortOn f = sortBy (\x y -> compare (f x) (f y))