BUG Fix parsing of GFF files

It was assumed that the score field was always non-negative.

Reported by Josh Sekela on the mailing-list:
https://groups.google.com/g/ngless/c/kf6y2MWBfec/m/2DicyAH3DwAJ

Author: luispedro

ChangeLog

@@ -1,3 +1,6 @@
+Version 1.4.2 2022-07-21 by luispedro
+	* Fix bug with parsing GFF files
+
 Version 1.4.1 2022-06-03 by luispedro
 	* Fix bug with split mapping
 	* Fix packaging for hackage

NGLess/Data/GFF.hs

@@ -1,4 +1,4 @@
-{- Copyright 2013-2019 NGLess Authors
+{- Copyright 2013-2022 NGLess Authors
  - License: MIT -}
 {-# LANGUAGE CPP #-}
 
@@ -81,31 +81,32 @@ readGffLine :: B.ByteString -> Either NGError GffLine
 readGffLine line = case B8.split '\t' line of
         [tk0,tk1,tk2,tk3,tk4,tk5,tk6,tk7,tk8] ->
             GffLine
-                tk0
-                tk1
-                tk2
-                <$> intOrError tk3
-                <*> intOrError tk4
-                <*> score tk5
+                tk0 -- seq id
+                tk1 -- source
+                tk2 -- type
+                <$> intOrError "reading start" tk3 -- start
+                <*> intOrError "reading end" tk4 -- end
+                <*> score tk5 -- score
                 <*> strandOrError tk6
                 <*> phase tk7
                 <*> pure (_parseGffAttributes tk8)
         _ -> throwDataError ("unexpected line in GFF: " ++ show line)
     where
-        parseOrError :: (a -> Maybe b) -> a -> NGLess b
-        parseOrError p s = case p s of
+        parseOrError :: String -> (B.ByteString -> Maybe b) -> B.ByteString -> NGLess b
+        parseOrError context p s = case p s of
                     Just v -> return v
-                    Nothing -> throwDataError $ "Could not parse GFF line: "++ show line
-        intOrError :: B.ByteString -> NGLess Int
-        intOrError = parseOrError (liftM fst . I.readDecimal)
-        floatOrError = parseOrError (liftM fst . F.readDecimal)
+                    Nothing -> throwDataError $ "Could not parse GFF line (" ++ context  ++ ", while parsing '" ++ B8.unpack s ++  "'): "++ show line
+        intOrError :: String -> B.ByteString -> NGLess Int
+        intOrError c = parseOrError c (liftM fst . I.readDecimal)
+        floatOrError c = parseOrError c (liftM fst . (F.readSigned F.readDecimal))
+
         score :: B.ByteString -> NGLess (Maybe Float)
         score "." = return Nothing
-        score v = Just <$> floatOrError v
+        score v = Just <$> floatOrError "reading score" v
 
         phase :: B.ByteString -> NGLess Int
         phase "." = return (-1)
-        phase r = intOrError r
+        phase r = intOrError "reading phase" r
         strandOrError :: B.ByteString -> NGLess GffStrand
         strandOrError s = case B8.uncons s of
             Just (s',_) -> parseStrand s'

NGLess/Interpretation/Count.hs

@@ -792,10 +792,23 @@ loadGFF gffFp opts = do
         outputListLno' TraceOutput ["Loading GFF file '", gffFp, "'..."]
         numCapabilities <- liftIO getNumCapabilities
         let mapthreads = max 1 (numCapabilities - 1)
+            annotateErrorReader :: (Int, V.Vector ByteLine) -> NGLess (V.Vector GffLine)
+            annotateErrorReader (ch_ix, ells) =
+                case V.mapM (readGffLine. unwrapByteLine) . V.filter (not . isComment) $ ells of
+                    r@Right{} -> r
+                    _ -> do
+                        forM_ (zip [0..] $ V.toList ells) $ \(i, ell) ->
+                            if isComment ell
+                                then return ()
+                                else case readGffLine (unwrapByteLine ell) of
+                                        Right{} -> return ()
+                                        Left (NGError errtype errmsg) -> Left (NGError errtype (errmsg ++ " (Line " ++ show (8192 * ch_ix + i + 1) ++ ")"))
+                        throwShouldNotOccur "annotateErrorReader: this should never happen"
         partials <- C.runConduit $
                 conduitPossiblyCompressedFile gffFp
                     .| linesVC 8192
-                    .| CAlg.asyncMapEitherC mapthreads (V.mapM (readGffLine . unwrapByteLine) . V.filter (not . isComment))
+                    .| CAlg.enumerateC
+                    .| CAlg.asyncMapEitherC mapthreads annotateErrorReader
                     .| sequenceSinks
                         [CL.foldM (insertgV f sf) (GffLoadingState M.empty M.empty)
                                     |  f <- optFeatures    opts

docs/sources/whatsnew.rst

@@ -2,6 +2,16 @@
 What's New (History)
 ====================
 
+Version 1.4.2
+-------------
+
+Released *21 July 2022*
+
+Bugfixes
+~~~~~~~~
+
+- Fix bug with parsing GFF files (it was assumed that _scores_ were always positive)
+
 Version 1.4.1
 -------------
 

test_samples/short.gtf

@@ -1885,3 +1885,5 @@ X	protein_coding	CDS	12148155	12148247	.	+	0	gene_id "WBGene00010924"; transcrip
 X	protein_coding	stop_codon	12148248	12148250	.	+	0	gene_id "WBGene00010924"; transcript_id "M153.1"; exon_number "5"; gene_name "M153.1"; gene_source "ensembl"; gene_biotype "protein_coding"; transcript_name "M153.1"; transcript_source "ensembl";
 X	protein_coding	UTR	12146817	12146844	.	+	.	gene_id "WBGene00010924"; transcript_id "M153.1"; gene_name "M153.1"; gene_source "ensembl"; gene_biotype "protein_coding"; transcript_name "M153.1"; transcript_source "ensembl";
 X	protein_coding	UTR	12148251	12148471	.	+	.	gene_id "WBGene00010924"; transcript_id "M153.1"; gene_name "M153.1"; gene_source "ensembl"; gene_biotype "protein_coding"; transcript_name "M153.1"; transcript_source "ensembl";
+k141_708	Prodigal_v2.6.3	CDS	1573	2394	46.2	-	0	ID=702_5;partial=00;start_type=ATG;rbs_motif=GGAG/GAGG;rbs_spacer=5-10bp;gc_cont=0.482;conf=100.00;score=45.55;cscore=32.79;sscore=12.76;rscore=2.32;uscore=7.11;tscore=3.98;
+k141_708	Prodigal_v2.6.3	CDS	1451	1576	-0.7	-	0	ID=702_4;partial=00;start_type=ATG;rbs_motif=None;rbs_spacer=None;gc_cont=0.444;conf=55.61;score=0.98;cscore=-3.53;sscore=4.51;rscore=-0.37;uscore=1.76;tscore=1.96;