"bufio"
"bytes"
"io"
- "log"
+ "strings"
"sync"
+ log "github.com/sirupsen/logrus"
"golang.org/x/crypto/blake2b"
)
-type tileVariantID int32 // 1-based
+type tileVariantID uint16 // 1-based
type tileLibRef struct {
tag tagID
type tileSeq map[string][]tileLibRef
+func (tseq tileSeq) Variants() []tileVariantID {
+ maxtag := 0
+ for _, refs := range tseq {
+ for _, ref := range refs {
+ if maxtag < int(ref.tag) {
+ maxtag = int(ref.tag)
+ }
+ }
+ }
+ vars := make([]tileVariantID, maxtag+1)
+ for _, refs := range tseq {
+ for _, ref := range refs {
+ vars[int(ref.tag)] = ref.variant
+ }
+ }
+ return vars
+}
+
type tileLibrary struct {
taglib *tagLibrary
- variant [][][blake2b.Size]byte
+ variant [][][blake2b.Size256]byte
// count [][]int
// seq map[[blake2b.Size]byte][]byte
variants int
if len(buf) == 0 || buf[0] == '>' {
todo <- jobT{seqlabel, fasta}
seqlabel, fasta = string(buf[1:]), nil
- log.Printf("%s %s reading fasta", filelabel, seqlabel)
+ log.Debugf("%s %s reading fasta", filelabel, seqlabel)
} else {
fasta = append(fasta, bytes.ToLower(buf)...)
}
}
todo <- jobT{seqlabel, fasta}
}()
+ path := make([]tileLibRef, 2000000)
+ totalPathLen := 0
+ skippedSequences := 0
for job := range todo {
if len(job.fasta) == 0 {
continue
+ } else if strings.Contains(job.label, "_") {
+ skippedSequences++
+ continue
}
- log.Printf("%s %s tiling", filelabel, job.label)
- var path []tileLibRef
+ log.Debugf("%s %s tiling", filelabel, job.label)
+ path = path[:0]
tilestart := -1 // position in fasta of tile that ends here
tiletagid := tagID(-1) // tag id starting tile that ends here
- tilelib.taglib.FindAll(job.fasta, func(id tagID, pos int) {
+ tilelib.taglib.FindAll(job.fasta, func(id tagID, pos, taglen int) {
if tilestart >= 0 {
- path = append(path, tilelib.getRef(tiletagid, job.fasta[tilestart:pos]))
+ path = append(path, tilelib.getRef(tiletagid, job.fasta[tilestart:pos+taglen]))
}
tilestart = pos
tiletagid = id
if tiletagid >= 0 {
path = append(path, tilelib.getRef(tiletagid, job.fasta[tilestart:]))
}
- ret[job.label] = path
- log.Printf("%s %s tiled with path len %d", filelabel, job.label, len(path))
+ pathcopy := make([]tileLibRef, len(path))
+ copy(pathcopy, path)
+ ret[job.label] = pathcopy
+ log.Debugf("%s %s tiled with path len %d", filelabel, job.label, len(path))
+ totalPathLen += len(path)
}
+ log.Printf("%s tiled with total path len %d in %d sequences (skipped %d sequences with '_' in name)", filelabel, totalPathLen, len(ret), skippedSequences)
return ret, scanner.Err()
}
// tilelib.seq = map[[blake2b.Size]byte][]byte{}
// }
if tilelib.variant == nil {
- tilelib.variant = make([][][blake2b.Size]byte, tilelib.taglib.Len())
- }
- hash, err := blake2b.New(32, nil)
- if err != nil {
- panic(err)
- }
- _, err = hash.Write(seq)
- if err != nil {
- panic(err)
+ tilelib.variant = make([][][blake2b.Size256]byte, tilelib.taglib.Len())
}
- var seqhash [blake2b.Size]byte
- copy(seqhash[:], hash.Sum(nil))
+ seqhash := blake2b.Sum256(seq)
for i, varhash := range tilelib.variant[tag] {
if varhash == seqhash {
return tileLibRef{tag: tag, variant: tileVariantID(i + 1)}