tilelib.go

   1 package main
   2
   3 import (
   4         "bufio"
   5         "bytes"
   6         "io"
   7         "log"
   8         "sync"
   9
  10         "golang.org/x/crypto/blake2b"
  11 )
  12
  13 type tileVariantID int32 // 1-based
  14
  15 type tileLibRef struct {
  16         tag     tagID
  17         variant tileVariantID
  18 }
  19
  20 type tileSeq map[string][]tileLibRef
  21
  22 type tileLibrary struct {
  23         taglib  *tagLibrary
  24         variant [][][blake2b.Size]byte
  25         // count [][]int
  26         // seq map[[blake2b.Size]byte][]byte
  27
  28         mtx sync.Mutex
  29 }
  30
  31 func (tilelib *tileLibrary) TileFasta(filelabel string, rdr io.Reader) (tileSeq, error) {
  32         ret := tileSeq{}
  33         type jobT struct {
  34                 label string
  35                 fasta []byte
  36         }
  37         todo := make(chan jobT)
  38         scanner := bufio.NewScanner(rdr)
  39         go func() {
  40                 defer close(todo)
  41                 var fasta []byte
  42                 var seqlabel string
  43                 for scanner.Scan() {
  44                         buf := scanner.Bytes()
  45                         if len(buf) == 0 || buf[0] == '>' {
  46                                 todo <- jobT{seqlabel, fasta}
  47                                 seqlabel, fasta = string(buf[1:]), nil
  48                                 log.Printf("%s %s reading fasta", filelabel, seqlabel)
  49                         } else {
  50                                 fasta = append(fasta, bytes.ToLower(buf)...)
  51                         }
  52                 }
  53                 todo <- jobT{seqlabel, fasta}
  54         }()
  55         for job := range todo {
  56                 if len(job.fasta) == 0 {
  57                         continue
  58                 }
  59                 log.Printf("%s %s tiling", filelabel, job.label)
  60                 var path []tileLibRef
  61                 tilestart := -1        // position in fasta of tile that ends here
  62                 tiletagid := tagID(-1) // tag id starting tile that ends here
  63                 tilelib.taglib.FindAll(job.fasta, func(id tagID, pos int) {
  64                         if tilestart >= 0 {
  65                                 path = append(path, tilelib.getRef(tiletagid, job.fasta[tilestart:pos]))
  66                         }
  67                         tilestart = pos
  68                         tiletagid = id
  69                 })
  70                 if tiletagid >= 0 {
  71                         path = append(path, tilelib.getRef(tiletagid, job.fasta[tilestart:]))
  72                 }
  73                 ret[job.label] = path
  74                 log.Printf("%s %s tiled with path len %d", filelabel, job.label, len(path))
  75         }
  76         return ret, scanner.Err()
  77 }
  78
  79 // Return a tileLibRef for a tile with the given tag and sequence,
  80 // adding the sequence to the library if needed.
  81 func (tilelib *tileLibrary) getRef(tag tagID, seq []byte) tileLibRef {
  82         tilelib.mtx.Lock()
  83         defer tilelib.mtx.Unlock()
  84         // if tilelib.seq == nil {
  85         //      tilelib.seq = map[[blake2b.Size]byte][]byte{}
  86         // }
  87         if len(tilelib.variant) <= int(tag) {
  88                 tilelib.variant = append(tilelib.variant, make([][][blake2b.Size]byte, int(tag)-len(tilelib.variant)+1)...)
  89         }
  90         hash, err := blake2b.New(32, nil)
  91         if err != nil {
  92                 panic(err)
  93         }
  94         _, err = hash.Write(seq)
  95         if err != nil {
  96                 panic(err)
  97         }
  98         var seqhash [blake2b.Size]byte
  99         copy(seqhash[:], hash.Sum(nil))
 100         for i, varhash := range tilelib.variant[tag] {
 101                 if varhash == seqhash {
 102                         return tileLibRef{tag: tag, variant: tileVariantID(i + 1)}
 103                 }
 104         }
 105         tilelib.variant[tag] = append(tilelib.variant[tag], seqhash)
 106         // tilelib.seq[seqhash] = append([]byte(nil), seq...)
 107         return tileLibRef{tag: tag, variant: tileVariantID(len(tilelib.variant[tag]))}
 108 }