tilelib.go

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