# Difference between revisions of "Shootout/Fasta"

DonStewart (talk | contribs) (add trac number for new sjanssen/fasta) |
|||

Line 78: | Line 78: | ||

initSeed = nextSeed 42 |
initSeed = nextSeed 42 |
||

</haskell> |
</haskell> |
||

+ | |||

+ | [[User:ArthurVanLeeuwen|ArthurVanLeeuwen]] 13:38, 26 January 2007 (UTC) Changing chooseBase to not do the (p-f), adding |
||

+ | <haskell> |
||

+ | cumulative bfs = zip bases (scanl1 (+) frequencies) |
||

+ | where (bases,frequencies) = unzip bfs |
||

+ | </haskell> |
||

+ | and passing (cumulative iub) rather than iub (etc.) may give another 10 percent speedup. |
||

= Current best entry (Haskell GHC #3) = |
= Current best entry (Haskell GHC #3) = |

## Revision as of 13:38, 26 January 2007

This is a Shootout Entry for [1].

The description of the program:

Each program should

- encode the expected cumulative probabilities for 2 alphabets
- generate DNA sequences, by weighted random selection from the alphabets (using the pseudo-random number generator from the random benchmark)
- generate DNA sequences, by copying from a given sequence
- write 3 sequences line-by-line in FASTA format

We'll use the generated FASTA file as input for other benchmarks (reverse-complement, k-nucleotide).

# Proposed entry

This entry is similar to the "Old entry" on this page, but has been updated to use Data.ByteString. It appears to be respectably fast, execution time is about 3.5 times the C++ version in initial tests.

It has been submitted.

```
{- The Computer Language Shootout
http://shootout.alioth.debian.org/
contributed by Jeff Newbern
updated by Spencer Janssen and Don Stewart -}
import System
import qualified Data.ByteString.Char8 as B
randomSequence :: Int -> [(Char,Double)] -> Int -> (B.ByteString, Int)
randomSequence n bf seed = (sequence, seed')
where (sequence, Just seed') = B.unfoldrN n f seed
f s = Just (chooseBase bf (normalize s), nextSeed s)
chooseBase :: [(Char,Double)] -> Double -> Char
chooseBase [(b,_)] _ = b
chooseBase ((b,f):xs) p | p < f = b
| otherwise = chooseBase xs (p-f)
writeFasta label title sequence = do
putStrLn $ ">" ++ label ++ " " ++ title
mapM_ B.putStrLn $ split 60 sequence
split n xs | B.null xs = []
| otherwise = l : split n r
where (l, r) = B.splitAt n xs
main = do
n <- getArgs >>= readIO . head
let aluLen = 1 + 2 * n `div` B.length alu
aluSeq = B.take (2 * n) . B.concat . replicate aluLen $ alu
(iubSeq, seed) = randomSequence (3 * n) iub initSeed
(homSeq, _) = randomSequence (5 * n) homosapiens seed
writeFasta "ONE" "Homo sapiens alu" aluSeq
writeFasta "TWO" "IUB ambiguity codes" iubSeq
writeFasta "THREE" "Homo sapiens frequency" homSeq
alu = B.pack
"GGCCGGGCGCGGTGGCTCACGCCTGTAATCCCAGCACTTTGG\
\GAGGCCGAGGCGGGCGGATCACCTGAGGTCAGGAGTTCGAGA\
\CCAGCCTGGCCAACATGGTGAAACCCCGTCTCTACTAAAAAT\
\ACAAAAATTAGCCGGGCGTGGTGGCGCGCGCCTGTAATCCCA\
\GCTACTCGGGAGGCTGAGGCAGGAGAATCGCTTGAACCCGGG\
\AGGCGGAGGTTGCAGTGAGCCGAGATCGCGCCACTGCACTCC\
\AGCCTGGGCGACAGAGCGAGACTCCGTCTCAAAAA"
iub = [ ('a', 0.27), ('c', 0.12), ('g', 0.12), ('t', 0.27), ('B', 0.02)
, ('D', 0.02), ('H', 0.02), ('K', 0.02), ('M', 0.02), ('N', 0.02)
, ('R', 0.02), ('S', 0.02), ('V', 0.02), ('W', 0.02), ('Y', 0.02) ]
homosapiens = [ ('a', 0.3029549426680), ('c', 0.1979883004921),
('g', 0.1975473066391), ('t', 0.3015094502008) ]
im = 139968
ia = 3877
ic = 29573
nextSeed s = (s * ia + ic) `rem` im
normalize n = (fromIntegral n) * (1.0 / fromIntegral im)
initSeed = nextSeed 42
```

ArthurVanLeeuwen 13:38, 26 January 2007 (UTC) Changing chooseBase to not do the (p-f), adding

```
cumulative bfs = zip bases (scanl1 (+) frequencies)
where (bases,frequencies) = unzip bfs
```

and passing (cumulative iub) rather than iub (etc.) may give another 10 percent speedup.

# Current best entry (Haskell GHC #3)

This entry has been disqualified. From the Shootout judges: "Preconverting to Int is a really interesting approach - we're looking for the vanilla approach."

It now has respectable performance and memory usage.

```
{-# OPTIONS_GHC -O2 -funbox-strict-fields #-}
-- The Great Computer Language Shootout
-- http://shootout.alioth.debian.org/
--
-- contributed by Jeff Newbern
-- Modified to fastest.hs by Chris Kuklewicz, 6 Jan 2006
-- Modified to fixed-fasta.hs by Chris Kuklewicz, 17 Jan 2006
--
-- Uses random generation code derived from Simon Marlow and Einar
-- Karttunen's "random" test entry. No longer uses Double during run,
-- everything has been pre-converted to Int. And pre-converted to a
-- binary tree for lookup. Ideally this tree could be constructed
-- with the probabilities in mind, but it isn't in this version.
--
-- Compile with ghc --make resub-fasta.hs -o resub-fasta.ghc_run
-- Run with "./rsub-fasta.ghc_run %A" where %A is the parameter
import Control.Monad
import Data.Char(chr,ord)
import Data.List(mapAccumL)
import Data.Word(Word8)
import Data.IORef
import Foreign
import System(getArgs)
import System.IO
type Base = Word8
data BaseFrequencyTree = Node !Base
| TreeNodes !Int !Base !Base
| Tree !Int !BaseFrequencyTree !BaseFrequencyTree
data Seed = Seed !Int
b2c :: Word8 -> Char
b2c = chr . fromEnum
c2b = toEnum . ord
alu = "GGCCGGGCGCGGTGGCTCACGCCTGTAATCCCAGCACTTTGG" ++
"GAGGCCGAGGCGGGCGGATCACCTGAGGTCAGGAGTTCGAGA" ++
"CCAGCCTGGCCAACATGGTGAAACCCCGTCTCTACTAAAAAT" ++
"ACAAAAATTAGCCGGGCGTGGTGGCGCGCGCCTGTAATCCCA" ++
"GCTACTCGGGAGGCTGAGGCAGGAGAATCGCTTGAACCCGGG" ++
"AGGCGGAGGTTGCAGTGAGCCGAGATCGCGCCACTGCACTCC" ++
"AGCCTGGGCGACAGAGCGAGACTCCGTCTCAAAAA"
im = 139968 :: Double
iub = mkTree $ snd . mapAccumL (\rt (c,f) -> (f+rt,(c2b c,ceiling $ im*(f+rt)))) 0.0 $
[ ('a', 0.27), ('c', 0.12), ('g', 0.12), ('t', 0.27), ('B', 0.02),
('D', 0.02), ('H', 0.02), ('K', 0.02), ('M', 0.02), ('N', 0.02),
('R', 0.02), ('S', 0.02), ('V', 0.02), ('W', 0.02), ('Y', 0.02) ]
homosapiens = mkTree $ snd . mapAccumL (\rt (c,f) -> (f+rt,(c2b c,ceiling $ im*(f+rt)))) 0.0 $
[ ('a', 0.3029549426680), ('c', 0.1979883004921), ('g', 0.1975473066391), ('t', 0.3015094502008) ]
mkTree [(b,_)] = Node b
mkTree [(b,f),(b',_)] = TreeNodes f b b'
mkTree xs = let (h,t) = splitAt (length xs `div` 2) xs
(_,f) = last h
in Tree f (mkTree h) (mkTree t)
chooseBase (Node b) _ = b
chooseBase (TreeNodes f b b') p = if (p<f) then b else b'
chooseBase (Tree f l r) p | p < f = chooseBase l p
| otherwise = chooseBase r p
writeFastaHeader label title = (putStrLn (('>':label) ++ (' ':title)))
perLine = 60
writeAluBuffer total = do
let l = length alu
bufSize = l + perLine - 1
aluBuf <- mallocArray bufSize
foldM_ (\ptr c -> poke ptr (c2b c) >> return (advancePtr ptr 1)) aluBuf (take bufSize (cycle alu))
let (full,end) = total `divMod` perLine
fullLine n = let ptr = advancePtr aluBuf ((n * perLine) `mod` l)
in hPutBuf stdout ptr perLine >> hPutChar stdout '\n'
lastLine = let ptr = advancePtr aluBuf ((full*perLine) `mod` l)
in hPutBuf stdout ptr end >> hPutChar stdout '\n'
mapM_ fullLine [0..pred full]
when (end>0) lastLine
writeWrapped total trans initSeed = do
seedRef <- newIORef initSeed
let l = succ perLine
(im,ia,ic)=(139968,3877,29573)
nextSeed (Seed s) = Seed ( (s * ia + ic) `mod` im )
prng = do newSeed <- return.nextSeed =<< readIORef seedRef
writeIORef seedRef newSeed
return newSeed
buf <- mallocArray l
poke (advancePtr buf perLine) (c2b '\n')
let (full,end) = total `divMod` perLine
fill 0 _ = return ()
fill i ptr = do (Seed b) <- prng
poke ptr (trans b)
fill (pred i) (advancePtr ptr 1)
fullLine = do fill perLine buf
hPutBuf stdout buf l
lastLine = do fill end buf
poke (advancePtr buf end) (c2b '\n')
hPutBuf stdout buf (succ end)
replicateM_ full fullLine
when (end>0) lastLine
readIORef seedRef
main = do args <- getArgs
let n = if null args then 2500000 else read (head args)
writeFastaHeader "ONE" "Homo sapiens alu"
writeAluBuffer (2*n)
writeFastaHeader "TWO" "IUB ambiguity codes"
seed' <- writeWrapped (3*n) (chooseBase iub) (Seed 42)
writeFastaHeader "THREE" "Homo sapiens frequency"
writeWrapped (5*n) (chooseBase homosapiens) seed'
```

An extremely minor gripe, how about using:

```
alu = "GGCCGGGCGCGGTGGCTCACGCCTGTAATCCCAGCACTTTGG\
\GAGGCCGAGGCGGGCGGATCACCTGAGGTCAGGAGTTCGAGA\
\CCAGCCTGGCCAACATGGTGAAACCCCGTCTCTACTAAAAAT\
\ACAAAAATTAGCCGGGCGTGGTGGCGCGCGCCTGTAATCCCA\
\GCTACTCGGGAGGCTGAGGCAGGAGAATCGCTTGAACCCGGG\
\AGGCGGAGGTTGCAGTGAGCCGAGATCGCGCCACTGCACTCC\
\AGCCTGGGCGACAGAGCGAGACTCCGTCTCAAAAA"
```

Rather than all those ++... This won't help performance one bit I would wager, but uses the "proper" way of doing multiline strings.--SebastianSylvan 11:40, 10 November 2006 (UTC)

# Non-leaking Entry

This a new entry, which is mostly the old entry below modified to use StateT to allow the random number stream to be finite, so no splitAt or drop functions are needed. This fixes the space leak and improves the time by almost a factor of 2. Serious speed will require changing from String to a Word8 array of some sort.

```
{-
The Great Computer Language Shootout
http://shootout.alioth.debian.org/
contributed by Jeff Newbern
Modified by Chris Kuklewicz, 3 Jan 2006
Uses random generation code derived from Simon Marlow and Einar
Karttunen's "random" test entry.
Modified version note: Use a StateT around IO to manage
the seed. This makes interleaving the generation and output of the
sequences easier.
-}
import Control.Monad.Trans
import Control.Monad.State
import System(getArgs)
type Base = Char
type Sequence = [Base]
alu :: Sequence -- predefined sequence
alu = "GGCCGGGCGCGGTGGCTCACGCCTGTAATCCCAGCACTTTGG" ++
"GAGGCCGAGGCGGGCGGATCACCTGAGGTCAGGAGTTCGAGA" ++
"CCAGCCTGGCCAACATGGTGAAACCCCGTCTCTACTAAAAAT" ++
"ACAAAAATTAGCCGGGCGTGGTGGCGCGCGCCTGTAATCCCA" ++
"GCTACTCGGGAGGCTGAGGCAGGAGAATCGCTTGAACCCGGG" ++
"AGGCGGAGGTTGCAGTGAGCCGAGATCGCGCCACTGCACTCC" ++
"AGCCTGGGCGACAGAGCGAGACTCCGTCTCAAAAA"
type BaseFrequency = (Base,Double)
iub :: [BaseFrequency]
iub = [ ('a', 0.27), ('c', 0.12), ('g', 0.12), ('t', 0.27),
('B', 0.02), ('D', 0.02), ('H', 0.02), ('K', 0.02), ('M', 0.02), ('N', 0.02),
('R', 0.02), ('S', 0.02), ('V', 0.02), ('W', 0.02), ('Y', 0.02) ]
homosapiens :: [BaseFrequency]
homosapiens = [ ('a', 0.3029549426680), ('c', 0.1979883004921),
('g', 0.1975473066391), ('t', 0.3015094502008) ]
-- select a base whose interval covers the given double
chooseBase :: [BaseFrequency] -> Double -> Base
chooseBase [(b,_)] _ = b
chooseBase ((b,f):xs) p | p < f = b
| otherwise = chooseBase xs (p-f)
type Seed = Int
type Pseudo a = StateT Seed IO a
prng :: Pseudo Double
prng = let nextSeed s = (s * ia + ic) `mod` im
normalize n = (fromIntegral n) * (1.0 / fromIntegral im)
im = 139968; ia = 3877; ic = 29573
in do seed <- get
let seed' = nextSeed seed
put seed'
return (normalize seed')
prngN count = sequence $ replicate count prng
-- write a sequence in Fasta format
writeFasta :: String -> String -> Sequence -> IO ()
writeFasta label title sequence =
do putStrLn $ ">" ++ (label ++ (" " ++ title))
writeWrapped 60 sequence
where writeWrapped _ [] = do return ()
writeWrapped len str = do let (s1,s2) = splitAt len str
putStrLn s1
writeWrapped len s2
writeFastaHeader :: (MonadIO m) => String -> String -> m ()
writeFastaHeader label title = liftIO $ putStrLn $ ">" ++ (label ++ (" " ++ title))
writeWrapped' :: Int -> Int -> (Double->Base) -> Pseudo ()
writeWrapped' wrap total trans =
let work c = case c of
0 -> return ()
n -> do let c' = min wrap n
nextC = c - c'
s <- liftM (map trans) (prngN c')
liftIO $ putStrLn s
work nextC
in work total
writeWrapped = writeWrapped' 60
main = do args <- getArgs
let n = if (null args) then 1000 else read (head args)
writeFasta "ONE" "Homo sapiens alu" (take (2*n) (cycle alu))
writeFastaHeader "TWO" "IUB ambiguity codes"
seed' <- execStateT (writeWrapped (3*n) (chooseBase iub)) 42
writeFastaHeader "THREE" "Homo sapiens frequency"
execStateT (writeWrapped (5*n) (chooseBase homosapiens)) seed'
```

# Old Entry

This is the old entry. It has a huge memory leak due to laziness.

```
-- The Great Computer Language Shootout
-- http://shootout.alioth.debian.org/
-- contributed by Jeff Newbern
-- Uses random generation code derived from Simon Marlow and Einar Karttunen's
-- "random" test entry.
-- Note: This code has not been optimized *at all*. It is written to be clear
-- and to follow standard Haskell idioms as much as possible (but we have to match
-- the stateful PRNG idiom in the test definition, which is oriented toward an
-- imperative language). Performance is decent with ghc -O2, but if it can be
-- improved without sacrificing the clarity of the code, by all means go for it!
import System(getArgs)
type Base = Char
type Sequence = [Base]
alu :: Sequence -- predefined sequence
alu = "GGCCGGGCGCGGTGGCTCACGCCTGTAATCCCAGCACTTTGG" ++
"GAGGCCGAGGCGGGCGGATCACCTGAGGTCAGGAGTTCGAGA" ++
"CCAGCCTGGCCAACATGGTGAAACCCCGTCTCTACTAAAAAT" ++
"ACAAAAATTAGCCGGGCGTGGTGGCGCGCGCCTGTAATCCCA" ++
"GCTACTCGGGAGGCTGAGGCAGGAGAATCGCTTGAACCCGGG" ++
"AGGCGGAGGTTGCAGTGAGCCGAGATCGCGCCACTGCACTCC" ++
"AGCCTGGGCGACAGAGCGAGACTCCGTCTCAAAAA"
type BaseFrequency = (Base,Double)
iub :: [BaseFrequency]
iub = [ ('a', 0.27), ('c', 0.12), ('g', 0.12), ('t', 0.27),
('B', 0.02), ('D', 0.02), ('H', 0.02), ('K', 0.02), ('M', 0.02), ('N', 0.02),
('R', 0.02), ('S', 0.02), ('V', 0.02), ('W', 0.02), ('Y', 0.02) ]
homosapiens :: [BaseFrequency]
homosapiens = [ ('a', 0.3029549426680), ('c', 0.1979883004921),
('g', 0.1975473066391), ('t', 0.3015094502008) ]
-- select a base whose interval covers the given double
chooseBase :: [BaseFrequency] -> Double -> Base
chooseBase [(b,_)] _ = b
chooseBase ((b,f):xs) p | p < f = b
| otherwise = chooseBase xs (p-f)
-- write a sequence in Fasta format
writeFasta :: String -> String -> Sequence -> IO ()
writeFasta label title sequence =
do putStrLn $ ">" ++ (label ++ (" " ++ title))
writeWrapped 60 sequence
where writeWrapped _ [] = do return ()
writeWrapped len str = do let (s1,s2) = splitAt len str
putStrLn s1
writeWrapped len s2
-- generate an infinite sequence of random doubles using the
-- prng from the "random" test
probs :: Int -> [Double]
probs seed = tail $ map normalize (iterate nextSeed seed)
where nextSeed s = (s * ia + ic) `mod` im
im = 139968
ia = 3877
ic = 29573
normalize n = (fromIntegral n) * (1.0 / fromIntegral im)
main = do args <- getArgs
let n = if (null args) then 1000 else read (head args)
writeFasta "ONE" "Homo sapiens alu" (take (2*n) (cycle alu))
let (seq1,seq2) = splitAt (3*n) (probs 42) -- we have to match the imperative version
let seq2' = take (5*n) seq2 -- instead of using the Haskell idiom
writeFasta "TWO" "IUB ambiguity codes" (map (chooseBase iub) seq1)
writeFasta "THREE" "Homo sapiens frequency" (map (chooseBase homosapiens) seq2')
```