X-Git-Url: https://git.arvados.org/lightning.git/blobdiff_plain/00b2acd54dd1aa412f6f2bddc24b1bbb31c7ae3f..124f6eee3729446b0e095f788f81590ee1dc880d:/annotate.go

diff --git a/annotate.go b/annotate.go
index ac34ada40f..25651889aa 100644
--- a/annotate.go
+++ b/annotate.go
@@ -1,3 +1,7 @@
+// Copyright (C) The Lightning Authors. All rights reserved.
+//
+// SPDX-License-Identifier: AGPL-3.0
+
 package lightning
 
 import (
@@ -23,9 +27,11 @@ import (
 )
 
 type annotatecmd struct {
-	variantHash bool
-	maxTileSize int
-	tag2tagid   map[string]tagID
+	dropTiles        []bool
+	variantHash      bool
+	maxTileSize      int
+	tag2tagid        map[string]tagID
+	reportAnnotation func(tag tagID, outcol int, variant tileVariantID, refname string, seqname string, pdi hgvs.Variant)
 }
 
 func (cmd *annotatecmd) RunCommand(prog string, args []string, stdin io.Reader, stdout, stderr io.Writer) int {
@@ -112,7 +118,7 @@ func (cmd *annotatecmd) RunCommand(prog string, args []string, stdin io.Reader,
 		retainNoCalls:       true,
 		retainTileSequences: true,
 	}
-	err = tilelib.LoadGob(context.Background(), input, strings.HasSuffix(*inputFilename, ".gz"), nil)
+	err = tilelib.LoadGob(context.Background(), input, strings.HasSuffix(*inputFilename, ".gz"))
 	if err != nil {
 		return 1
 	}
@@ -183,6 +189,9 @@ func (cmd *annotatecmd) exportTileDiffs(outw io.Writer, tilelib *tileLibrary) er
 		for _, seqname := range seqnames {
 			seqname := seqname
 			throttle.Acquire()
+			if throttle.Err() != nil {
+				break
+			}
 			go func() {
 				defer throttle.Release()
 				throttle.Report(cmd.annotateSequence(throttle, outch, tilelib, taglen, refname, seqname, refcs[seqname], len(refs) > 1))
@@ -194,6 +203,10 @@ func (cmd *annotatecmd) exportTileDiffs(outw io.Writer, tilelib *tileLibrary) er
 }
 
 func (cmd *annotatecmd) annotateSequence(throttle *throttle, outch chan<- string, tilelib *tileLibrary, taglen int, refname, seqname string, reftiles []tileLibRef, refnamecol bool) error {
+	refnamefield := ""
+	if refnamecol {
+		refnamefield = "," + trimFilenameForLabel(refname)
+	}
 	var refseq []byte
 	// tilestart[123] is the index into refseq
 	// where the tile for tag 123 was placed.
@@ -216,17 +229,29 @@ func (cmd *annotatecmd) annotateSequence(throttle *throttle, outch chan<- string
 		tileend[libref.Tag] = len(refseq)
 	}
 	log.Infof("seq %s len(refseq) %d len(tilestart) %d", seqname, len(refseq), len(tilestart))
+	// outtag is tag's index in the subset of tags that aren't
+	// dropped. If there are 10M tags and half are dropped by
+	// dropTiles, tag ranges from 0 to 10M-1 and outtag ranges
+	// from 0 to 5M-1.
+	//
+	// IOW, in the matrix built by cgs2array(), {tag} is
+	// represented by columns {outtag}*2 and {outtag}*2+1.
+	outcol := -1
 	for tag, tvs := range tilelib.variant {
+		if len(cmd.dropTiles) > tag && cmd.dropTiles[tag] {
+			continue
+		}
 		tag := tagID(tag)
+		outcol++
+		// Must shadow outcol var to use safely in goroutine below.
+		outcol := outcol
 		refstart, ok := tilestart[tag]
 		if !ok {
-			// Tag didn't place on this
-			// reference sequence. (It
-			// might place on the same
-			// chromosome in a genome
-			// anyway, but we don't output
-			// the annotations that would
-			// result.)
+			// Tag didn't place on this reference
+			// sequence. (It might place on the same
+			// chromosome in a genome anyway, but we don't
+			// output the annotations that would result.)
+			// outch <- fmt.Sprintf("%d,%d,-1%s\n", tag, outcol, refnamefield)
 			continue
 		}
 		for variant := 1; variant <= len(tvs); variant++ {
@@ -275,11 +300,10 @@ func (cmd *annotatecmd) annotateSequence(throttle *throttle, outch chan<- string
 					} else {
 						varid = fmt.Sprintf("%d", variant)
 					}
-					refnamefield := ""
-					if refnamecol {
-						refnamefield = "," + trimFilenameForLabel(refname)
+					outch <- fmt.Sprintf("%d,%d,%s%s,%s:g.%s\n", tag, outcol, varid, refnamefield, seqname, diff.String())
+					if cmd.reportAnnotation != nil {
+						cmd.reportAnnotation(tag, outcol, variant, refname, seqname, diff)
 					}
-					outch <- fmt.Sprintf("%d,%s%s,%s:g.%s\n", tag, varid, refnamefield, seqname, diff.String())
 				}
 			}()
 		}