normalize 0.3.0.1 → 0.3.1.0

raw patch · 4 files changed

+99/−14 lines, 4 filesdep +sparse-linear-algebra

Dependencies added: sparse-linear-algebra

Files

• normalize.cabal +12/−10
• src/Normalize.hs +26/−3
• src/Types.hs +6/−1
• src/Utility.hs +55/−0

@@ -1,13 +1,13 @@ name:                normalize-version:             0.3.0.1+version:             0.3.1.0 synopsis:            Normalize data using a variety of methods. description:         Normalize data using a variety of methods. For use with csv files. homepage:            http://github.com/GregorySchwartz/normalize#readme license:             GPL-3 license-file:        LICENSE author:              Gregory W. Schwartz-maintainer:          gsch@mail.med.upenn.edu-copyright:           Copyright: (c) 2017 Gregory W. Schwartz+maintainer:          gsch@pennmedicine.upenn.edu+copyright:           Copyright: (c) 2018 Gregory W. Schwartz category:            Bioinformatics build-type:          Simple -- extra-source-files:@@ -19,13 +19,15 @@                      , Load                      , Normalize                      , Filter+                     , Utility   build-depends:       base >= 4.7 && < 5+                     , cassava                      , containers                      , lens-                     , vector-                     , text-                     , cassava+                     , sparse-linear-algebra                      , statistics+                     , text+                     , vector   ghc-options:         -O2   default-language:    Haskell2010 @@ -35,12 +37,12 @@   ghc-options:         -threaded -rtsopts -O2   build-depends:       base                      , normalize-                     , optparse-generic-                     , containers-                     , vector                      , bytestring-                     , text                      , cassava+                     , containers+                     , optparse-generic+                     , text+                     , vector   default-language:    Haskell2010  source-repository head

@@ -25,14 +25,16 @@ import Data.Function (on)  -- Cabal-import qualified Data.Vector as V-import qualified Data.Text as T+import Control.Lens import Statistics.Quantile+import qualified Data.Sparse.Common as S+import qualified Data.Text as T+import qualified Data.Vector as V import qualified Statistics.Sample as Stat-import Control.Lens  -- Local import Types+import Utility  -- | Log transform the normalize map. logTransform :: Base -> Map.Map Sample (V.Vector Entity) -> Map.Map Sample (V.Vector Entity)@@ -52,7 +54,15 @@ normalize StandardScore = Map.map standardScore normalize UpperQuartile = Map.map upperQuartileNormalize normalize None          = id+normalize _             = error "Method not supported by normalize." +-- | Normalize all samples by a specific method using a sparse matrix.+normalizeSparse :: Method -> S.SpMatrix Double -> S.SpMatrix Double+normalizeSparse method@QuantileMedian  = quantileNormalize method+normalizeSparse method@QuantileAverage = quantileNormalize method+normalizeSparse None            = id+normalizeSparse _ = error "Method not supported by normalizeSparse."+ -- | Normalize a sample (1) by another sample (2) by division. The -- NormSampleString contains the string that differentiates (1) from (2). -- NormSampleString must be within (2) and must make, upon its removal from (2),@@ -145,3 +155,16 @@   where     zeroFiltered = V.filter ((> 0) . _value) xs     uqVal = continuousBy (ContParam 1 1) 3 4 . fmap _value++-- | Quantile normalization for sparse matrices, ignoring zeros.+quantileNormalize :: Method -> S.SpMatrix Double -> S.SpMatrix Double+quantileNormalize method mat =+    fmap (\x -> S.lookupDenseSV (x - 1) summaryVec) rankMat+  where+    summaryFunc QuantileMedian  = medianSparseVector+    summaryFunc QuantileAverage = avgSparseVector+    summaryFunc _ = error "Unsupported method for quantile normalization."+    summaryVec =+        S.sparsifySV . S.vr . fmap (summaryFunc method) . S.toRowsL $ sortMat+    sortMat    = S.fromColsL . fmap sortSparseVector . S.toColsL $ mat+    rankMat    = S.fromColsL . fmap rankSparseVector . S.toColsL $ mat

@@ -40,7 +40,12 @@ -- Advanced  -- Algebraic-data Method = StandardScore | UpperQuartile | None deriving (Eq, Read, Show)+data Method = StandardScore+            | UpperQuartile+            | QuantileMedian+            | QuantileAverage+            | None+            deriving (Eq, Read, Show)  data Entity = Entity                 { _label      :: !T.Text

@@ -0,0 +1,55 @@+{- Utility+Gregory W. Schwartz++Collections helper functions for the program.+-}++{-# LANGUAGE BangPatterns #-}++module Utility+    ( sortSparseVector+    , rankSparseVector+    , medianSparseVector+    , avgSparseVector+    ) where++-- Standard+import Data.Function (on)+import Data.List++-- Cabal+import Control.Lens+import Statistics.Quantile+import Statistics.Sample (mean)+import qualified Data.Map.Strict as Map+import qualified Data.Set as Set+import qualified Data.Sparse.Common as S+import qualified Data.Vector.Unboxed as V++-- Local++-- | Find the average of a sparse vector, ignoring zeros.+avgSparseVector :: S.SpVector Double -> Double+avgSparseVector xs = mean . V.fromList . fmap snd . S.toListSV $ xs++-- | Find the median of a sparse vector, ignoring zeros.+medianSparseVector :: S.SpVector Double -> Double+medianSparseVector xs =+    continuousBy s 2 4 . V.fromList . fmap snd . S.toListSV $ xs++-- | Sort a sparse vector, ignoring zeros.+sortSparseVector :: S.SpVector Double -> S.SpVector Double+sortSparseVector xs =+    S.fromListDenseSV (S.svDim xs) . sort . fmap snd . S.toListSV $ xs++-- | Get the rank transformed vector of a sparse vector, ignoring zeros.+rankSparseVector :: S.SpVector Double -> S.SpVector Int+rankSparseVector xs = fmap (\k -> Map.findWithDefault 0 k rankMap) xs+  where+    rankMap = Map.fromList+            . flip zip [1,2..]+            . Set.toList+            . Set.fromList+            . fmap snd+            . S.toListSV+            $ xs