Пример #1
0
func main() {
	giMapper, err := giTaxid.Load(dictfile, true)
	if err != nil {
		log.Fatalf("Problem reading dict file: %s\n", err)
	}

	taxid, err := giMapper.GiTaxid(gi)
	if err != nil {
		log.Fatalf("Problem retrieving taxid from dict: %s\n", err)
	}
	fmt.Printf("%d\n", taxid)
}
Пример #2
0
// New creates a new NCBI taxonomy representation from the options given in opts
// returns the newly created taxonomy or any error it may encounter in the process
func New(nodesfn, namesfn, dictfn string, savemem bool) (*Taxonomy, error) {
	// T : taxtree => tax
	t := &Taxonomy{}
	fmt.Fprintf(os.Stderr, "Creating new taxonomy tree ... ")
	s1 := time.Now()
	maxNodes, ewcl := wcl.FromFile(nodesfn)
	if ewcl != nil {
		return nil, ewcl
	}
	nodesf, eopen := os.OpenFile(nodesfn, os.O_RDONLY, 0644)
	if eopen != nil {
		return nil, eopen
	}
	nodesbuf := bufio.NewReader(io.Reader(nodesf))

	tax, dict, ee := newTaxonomy(nodesbuf, maxNodes)
	if ee != nil {
		return nil, ee
	}
	s2 := time.Now()
	dur := s2.Sub(s1)
	fmt.Fprintf(os.Stderr, "Done (%.3f sec)\n", dur.Seconds())

	// T.Name(s)
	fmt.Fprintf(os.Stderr, "Filling names in taxonomy tree ... ")
	s1 = time.Now()
	err := tax.loadNames(namesfn, dict)
	if err != nil {
		return nil, err
	}
	s2 = time.Now()
	dur = s2.Sub(s1)
	fmt.Fprintf(os.Stderr, "Done (%.3f sec)\n", dur.Seconds())

	// E, L, H
	fmt.Fprintf(os.Stderr, "Creating indexes ... ")
	s1 = time.Now()
	E := make([]int, 0, maxNodes*2)
	L := make([]int, 0, maxNodes*2)
	H := make([]int, maxNodes)
	tax.elh(1, 0, &E, &L, &H)
	s2 = time.Now()
	dur = s2.Sub(s1)
	fmt.Fprintf(os.Stderr, "Done (%.3f sec)\n", dur.Seconds())

	// M
	fmt.Fprintf(os.Stderr, "Preprocessing RMQ ... ")
	s1 = time.Now()
	M := makeMatrix(maxNodes)
	rmqPrep(&M, L[:len(L)-1], maxNodes)
	s2 = time.Now()
	dur = s2.Sub(s1)
	fmt.Fprintf(os.Stderr, "Done (%.3f sec)\n", dur.Seconds())

	fmt.Fprintf(os.Stderr, "Combining data ... ")
	s1 = time.Now()
	t.T = tax
	t.D = dict
	//		t.E = make([]int, len(E))
	//		copy(t.E, E)
	t.E = E
	//		t.L = make([]int, len(L))
	//		copy(t.L, L)
	t.L = L
	//		t.H = make([]int, len(H))
	//		copy(t.H, H)
	t.H = H
	// 		t.M = make([][]int, len(M))
	// 		for i := 0; i < len(M); i++ {
	// 			t.M[i] = make([]int, len(M[0]))
	// 		}
	// 		copy(t.M, M)
	t.M = M
	s2 = time.Now()
	dur = s2.Sub(s1)
	fmt.Fprintf(os.Stderr, "Done (%.3f sec)\n", dur.Seconds())

	t.G, err = giTaxid.Load(dictfn, savemem)
	if err != nil {
		return nil, err
	}
	return t, nil
}