biopsl (empty) → 0.1

raw patch · 4 files changed

+122/−0 lines, 4 filesdep +basedep +biocoredep +bytestringsetup-changed

Dependencies added: base, biocore, bytestring

Files

Setup.hs +3/−0
biopsl.cabal +25/−0
examples/PslUniq.hs +27/−0
src/Bio/Alignment/PSL.hs +67/−0

+ Setup.hs view

@@ -0,0 +1,3 @@+#!/usr/bin/env runhaskell+import Distribution.Simple+main = defaultMain

+ biopsl.cabal view

@@ -0,0 +1,25 @@+Name:                biopsl+Version:             0.1+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+		     code from (and is incompatible with) the "bio" library.+Homepage:            http://biohaskell.org/+License:             LGPL+Author:              Ketil Malde+Maintainer:          ketil@malde.org+Stability:           Experimental+Category:            Bioinformatics+Build-type:          Simple+Cabal-version:       >=1.2++Library+  Exposed-modules: Bio.Alignment.PSL+  Build-depends:   base >= 3 && < 5, biocore == 0.1, bytestring+  Hs-Source-Dirs:  src+  Ghc-Options:     -Wall++Executable psluniq+  Main-Is:        PslUniq.hs+  Hs-Source-Dirs: src, examples+  Ghc-Options:    -Wall

+ examples/PslUniq.hs view

@@ -0,0 +1,27 @@+{-| Select only best hit from each sequence from a PSL file, +    write to stdout +-}++import Bio.Alignment.PSL+import System.Environment (getArgs)+import qualified Data.ByteString.Lazy as L++main :: IO ()+main = do+  args <- getArgs+  ps <- case args of +        [] -> (parsePSL `fmap` L.getContents)+        [f] -> readPSL f+        _ -> error "Usage: psluniq [pslfile]"+  L.putStr pslHeader+  L.putStr $ unparsePSL $ uniq ps+  +uniq :: [PSL] -> [PSL]+uniq [] = []+uniq (p:ps) = 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 _ [] = []

+ src/Bio/Alignment/PSL.hs view

@@ -0,0 +1,67 @@+{-| This models the PSL format used by e.g. the alignment tool BLAT.  +    It is a simple, textual representation of (spliced) alignments,+    with tab-separated fields.++    See <http://genome.ucsc.edu/FAQ/FAQformat#format2> for details.+-}+module Bio.Alignment.PSL where++import qualified Data.ByteString.Lazy.Char8 as B+import Data.ByteString.Lazy.Char8 (ByteString)+import Data.List (intersperse)++-- | This encodes a PSL record, corresponding to one line of the PSL file.++-- NB! laziness issues, if not fully evaluated, this will hang onto the input.+data PSL = PSL   { match, mismatch, repmatch, ncount+                 , qgapcount, qgaplength, tgapcount, tgaplength :: Int +                 , strand :: ByteString+                 , qname :: ByteString, qsize, qstart, qend :: Int+                 , tname :: ByteString, tsize, tstart, tend :: Int+                 , blockcount :: Int, blocksizes, qstarts, tstarts :: [Int]+                 } deriving (Eq)++instance Show PSL where +  show = B.unpack . unparsePSL . return++-- | Read and parse a PSL file.+readPSL :: FilePath -> IO [PSL]+readPSL f = parsePSL `fmap` B.readFile f++-- | Create a PSL file from a list of alignments.+writePSL :: FilePath -> [PSL] -> IO ()+writePSL f =  B.writeFile f . B.append pslHeader . unparsePSL ++-- | Parse a 'ByteString' as a PSL file (note that it must contain the PSL header).+parsePSL :: ByteString -> [PSL]+parsePSL s = map parseLine $ B.lines $ dropHeader+  where+    -- dropHeader requires strict adherence to BLAT output.  Perhaps a looser check is better?+    dropHeader | pslHeader `B.isPrefixOf` s = B.drop (B.length pslHeader) s+               | otherwise = error "PSL header mismatch: couldn't determine the header."+    parseLine l = let fs = B.split '\t' l+                      ri :: Int -> Int+                      ri i = maybe (error ("Can't parse field "++show i++" '"++B.unpack (fs!!i)++"' as an integer")) fst $ B.readInt (fs!!i)+                      rl j = map (\w -> maybe undefined fst (B.readInt w)) $ init $ B.split ',' (fs!!j)+                  in PSL (ri 0) (ri 1) (ri 2) (ri 3) (ri 4) (ri 5) (ri 6) (ri 7) (fs!!8) (fs!!9) (ri 10) (ri 11) (ri 12) (fs!!13) (ri 14) (ri 15) (ri 16) (ri 17) (rl 18) (rl 19) (rl 20)+                           ++-- | Unparse a list of 'PSL' alignments encoding them into a 'ByteString' (not including PSL header).+unparsePSL :: [PSL] -> ByteString+unparsePSL = B.unlines . map format1+  where format1 :: PSL -> ByteString+        format1 (PSL m mm rm nc qc ql tgc tgl st qn qsz qst qed tn tsz tst ted bc bs qss tss) = +          B.concat $ intersperse tab $ (map showint [m, mm, rm, nc, qc, ql, tgc, tgl]+                                        ++[st,qn]++map showint [qsz,qst,qed]+                                        ++[tn]   ++map showint [tsz,tst,ted]+                                        ++[showint bc]++map showlist [bs,qss,tss])+        tab = B.pack "\t"+        showint  = B.pack . show+        showlist = B.pack . concat . map (++",") . map show++-- | The PSL header (version 3), as a 'ByteString'.+pslHeader :: ByteString+pslHeader = B.pack ("psLayout version 3\n\n"+                    ++"match\tmis- \trep. \tN's\tQ gap\tQ gap\tT gap\tT gap\tstrand\tQ        \tQ   \tQ    \tQ  \tT        \tT   \tT    \tT  \tblock\tblockSizes \tqStarts\t tStarts\n"+                    ++"     \tmatch\tmatch\t   \tcount\tbases\tcount\tbases\t      \tname     \tsize\tstart\tend\tname     \tsize\tstart\tend\tcount\n"+                    ++"---------------------------------------------------------------------------------------------------------------------------------------------------------------\n")