retainNoCalls bool
skipOOO bool
retainTileSequences bool
+ useDups bool
taglib *tagLibrary
variant [][][blake2b.Size256]byte
variants int64
// if non-nil, write out any tile variants added while tiling
encoder *gob.Encoder
+ // set Ref flag when writing new variants to encoder
+ encodeRef bool
onAddTileVariant func(libref tileLibRef, hash [blake2b.Size256]byte, seq []byte) error
onAddGenome func(CompactGenome) error
for _, tv := range tvs {
// Assign a new variant ID (unique across all inputs)
// for each input variant.
- variantmap[tileLibRef{Tag: tv.Tag, Variant: tv.Variant}] = tilelib.getRef(tv.Tag, tv.Sequence).Variant
+ variantmap[tileLibRef{Tag: tv.Tag, Variant: tv.Variant}] = tilelib.getRef(tv.Tag, tv.Sequence, tv.Ref).Variant
}
return nil
}
return nil
}
-func allGobFiles(path string) ([]string, error) {
+func allFiles(path string, re *regexp.Regexp) ([]string, error) {
var files []string
f, err := open(path)
if err != nil {
if fi.Name() == "." || fi.Name() == ".." {
continue
} else if child := path + "/" + fi.Name(); fi.IsDir() {
- add, err := allGobFiles(child)
+ add, err := allFiles(child, re)
if err != nil {
return nil, err
}
files = append(files, add...)
- } else if strings.HasSuffix(child, ".gob") || strings.HasSuffix(child, ".gob.gz") {
+ } else if re == nil || re.MatchString(child) {
files = append(files, child)
}
}
+ sort.Strings(files)
return files, nil
}
+var matchGobFile = regexp.MustCompile(`\.gob(\.gz)?$`)
+
func (tilelib *tileLibrary) LoadDir(ctx context.Context, path string) error {
log.Infof("LoadDir: walk dir %s", path)
- files, err := allGobFiles(path)
+ files, err := allFiles(path, matchGobFile)
if err != nil {
return err
}
mtx.Unlock()
}
for _, tv := range ent.TileVariants {
- variantmap[tileLibRef{Tag: tv.Tag, Variant: tv.Variant}] = tilelib.getRef(tv.Tag, tv.Sequence).Variant
+ variantmap[tileLibRef{Tag: tv.Tag, Variant: tv.Variant}] = tilelib.getRef(tv.Tag, tv.Sequence, tv.Ref).Variant
}
cgs = append(cgs, ent.CompactGenomes...)
cseqs = append(cseqs, ent.CompactSequences...)
}
type importStats struct {
- InputFile string
- InputLabel string
- InputLength int
- InputCoverage int
- PathLength int
- DroppedOutOfOrderTiles int
+ InputFile string
+ InputLabel string
+ InputLength int
+ InputCoverage int
+ PathLength int
+ DroppedRepeatedTags int
+ DroppedOutOfOrderTags int
}
-func (tilelib *tileLibrary) TileFasta(filelabel string, rdr io.Reader, matchChromosome *regexp.Regexp) (tileSeq, []importStats, error) {
+func (tilelib *tileLibrary) TileFasta(filelabel string, rdr io.Reader, matchChromosome *regexp.Regexp, isRef bool) (tileSeq, []importStats, error) {
ret := tileSeq{}
type jobT struct {
label string
}
todo := make(chan jobT, 1)
scanner := bufio.NewScanner(rdr)
+ scanner.Buffer(make([]byte, 256), 1<<29) // 512 MiB, in case fasta does not have line breaks
go func() {
defer close(todo)
var fasta []byte
log.Warnf("%s %s no tags found", filelabel, job.label)
}
- skipped := 0
+ droppedDup := 0
+ if !tilelib.useDups {
+ // Remove any tags that appeared more than once
+ dup := map[tagID]bool{}
+ for _, ft := range found {
+ _, dup[ft.tagid] = dup[ft.tagid]
+ }
+ dst := 0
+ for _, ft := range found {
+ if !dup[ft.tagid] {
+ found[dst] = ft
+ dst++
+ }
+ }
+ droppedDup = len(found) - dst
+ log.Infof("%s %s dropping %d non-unique tags", filelabel, job.label, droppedDup)
+ found = found[:dst]
+ }
+
+ droppedOOO := 0
if tilelib.skipOOO {
- log.Infof("%s %s keeping longest increasing subsequence", filelabel, job.label)
keep := longestIncreasingSubsequence(len(found), func(i int) int { return int(found[i].tagid) })
for i, x := range keep {
found[i] = found[x]
}
- skipped = len(found) - len(keep)
+ droppedOOO = len(found) - len(keep)
+ log.Infof("%s %s dropping %d out-of-order tags", filelabel, job.label, droppedOOO)
found = found[:len(keep)]
}
} else {
endpos = found[i+1].pos + taglen
}
- path[i] = tilelib.getRef(f.tagid, job.fasta[startpos:endpos])
+ path[i] = tilelib.getRef(f.tagid, job.fasta[startpos:endpos], isRef)
if countBases(job.fasta[startpos:endpos]) != endpos-startpos {
atomic.AddInt64(&lowquality, 1)
}
ret[job.label] = pathcopy
basesIn := countBases(job.fasta)
- log.Infof("%s %s fasta in %d coverage in %d path len %d low-quality %d skipped-out-of-order %d", filelabel, job.label, len(job.fasta), basesIn, len(path), lowquality, skipped)
+ log.Infof("%s %s fasta in %d coverage in %d path len %d low-quality %d", filelabel, job.label, len(job.fasta), basesIn, len(path), lowquality)
stats = append(stats, importStats{
- InputFile: filelabel,
- InputLabel: job.label,
- InputLength: len(job.fasta),
- InputCoverage: basesIn,
- PathLength: len(path),
- DroppedOutOfOrderTiles: skipped,
+ InputFile: filelabel,
+ InputLabel: job.label,
+ InputLength: len(job.fasta),
+ InputCoverage: basesIn,
+ PathLength: len(path),
+ DroppedOutOfOrderTags: droppedOOO,
+ DroppedRepeatedTags: droppedDup,
})
totalPathLen += len(path)
// Return a tileLibRef for a tile with the given tag and sequence,
// adding the sequence to the library if needed.
-func (tilelib *tileLibrary) getRef(tag tagID, seq []byte) tileLibRef {
+func (tilelib *tileLibrary) getRef(tag tagID, seq []byte, usedByRef bool) tileLibRef {
dropSeq := false
if !tilelib.retainNoCalls {
for _, b := range seq {
tilelib.encoder.Encode(LibraryEntry{
TileVariants: []TileVariant{{
Tag: tag,
+ Ref: usedByRef,
Variant: variant,
Blake2b: seqhash,
Sequence: saveSeq,