tilelib.go

   1 package main
   2
   3 import (
   4         "bufio"
   5         "bytes"
   6         "crypto/md5"
   7         "io"
   8         "log"
   9         "sync"
  10 )
  11
  12 type tileVariantID int32 // 1-based
  13
  14 type tileLibRef struct {
  15         tag     tagID
  16         variant tileVariantID
  17 }
  18
  19 type tileSeq map[string][]tileLibRef
  20
  21 type tileLibrary struct {
  22         taglib  *tagLibrary
  23         variant [][][md5.Size]byte
  24         // count [][]int
  25         // seq map[[md5.Size]byte][]byte
  26
  27         mtx sync.Mutex
  28 }
  29
  30 func (tilelib *tileLibrary) TileFasta(filelabel string, rdr io.Reader) (tileSeq, error) {
  31         ret := tileSeq{}
  32         var wg sync.WaitGroup
  33         flush := func(seqlabel string, fasta []byte) {
  34                 defer wg.Done()
  35                 var path []tileLibRef
  36                 if len(fasta) == 0 {
  37                         return
  38                 }
  39                 tilestart := -1        // position in fasta of tile that ends here
  40                 tiletagid := tagID(-1) // tag id starting tile that ends here
  41                 tilelib.taglib.FindAll(fasta, func(id tagID, pos int) {
  42                         if tilestart >= 0 {
  43                                 path = append(path, tilelib.getRef(tiletagid, fasta[tilestart:pos]))
  44                         }
  45                         tilestart = pos
  46                         tiletagid = id
  47                 })
  48                 if tiletagid >= 0 {
  49                         path = append(path, tilelib.getRef(tiletagid, fasta[tilestart:]))
  50                 }
  51                 ret[seqlabel] = path
  52                 log.Printf("%s %s tiled with path len %d", filelabel, seqlabel, len(path))
  53         }
  54         var fasta []byte
  55         var seqlabel string
  56         scanner := bufio.NewScanner(rdr)
  57         for scanner.Scan() {
  58                 buf := scanner.Bytes()
  59                 if len(buf) == 0 || buf[0] == '>' {
  60                         wg.Add(1)
  61                         go flush(seqlabel, fasta)
  62                         fasta = nil
  63                         seqlabel = string(buf[1:])
  64                 } else {
  65                         fasta = append(fasta, bytes.ToLower(buf)...)
  66                 }
  67         }
  68         if err := scanner.Err(); err != nil {
  69                 return nil, err
  70         }
  71         wg.Add(1)
  72         go flush(seqlabel, fasta)
  73         wg.Wait()
  74         return ret, nil
  75 }
  76
  77 // Return a tileLibRef for a tile with the given tag and sequence,
  78 // adding the sequence to the library if needed.
  79 func (tilelib *tileLibrary) getRef(tag tagID, seq []byte) tileLibRef {
  80         tilelib.mtx.Lock()
  81         defer tilelib.mtx.Unlock()
  82         // if tilelib.seq == nil {
  83         //      tilelib.seq = map[[md5.Size]byte][]byte{}
  84         // }
  85         if len(tilelib.variant) <= int(tag) {
  86                 tilelib.variant = append(tilelib.variant, make([][][md5.Size]byte, int(tag)-len(tilelib.variant)+1)...)
  87         }
  88         seqhash := md5.Sum(seq)
  89         for i, varhash := range tilelib.variant[tag] {
  90                 if varhash == seqhash {
  91                         return tileLibRef{tag: tag, variant: tileVariantID(i + 1)}
  92                 }
  93         }
  94         tilelib.variant[tag] = append(tilelib.variant[tag], seqhash)
  95         // tilelib.seq[seqhash] = append([]byte(nil), seq...)
  96         return tileLibRef{tag: tag, variant: tileVariantID(len(tilelib.variant[tag]))}
  97 }