Golang linear.NewSeq函数代码示例

func init() {
	var err error
	m, err = NewMulti("example multi",
		[]seq.Sequence{
			linear.NewSeq("example DNA 1", []alphabet.Letter("ACGCTGACTTGGTGCACGT"), alphabet.DNA),
			linear.NewSeq("example DNA 2", []alphabet.Letter("ACGGTGACCTGGCGCGCAT"), alphabet.DNA),
			linear.NewSeq("example DNA 3", []alphabet.Letter("ACGATGACGTGGCGCTCAT"), alphabet.DNA),
		},
		seq.DefaultConsensus)

	if err != nil {
		panic(err)
	}
}

func (s *S) TestReadFasta(c *check.C) {
	var (
		obtainN []string
		obtainS [][]alphabet.Letter
	)

	for _, fa := range fas {
		r := NewReader(bytes.NewBufferString(fa), linear.NewSeq("", nil, alphabet.Protein))
		for {
			if s, err := r.Read(); err != nil {
				if err == io.EOF {
					break
				} else {
					c.Fatalf("Failed to read %q: %s", fa, err)
				}
			} else {
				t := s.(*linear.Seq)
				header := t.Name()
				if desc := t.Description(); len(desc) > 0 {
					header += " " + desc
				}
				obtainN = append(obtainN, header)
				obtainS = append(obtainS, t.Slice().(alphabet.Letters))
			}
		}
		c.Check(obtainN, check.DeepEquals, expectN)
		obtainN = nil
		for i := range obtainS {
			c.Check(len(obtainS[i]), check.Equals, len(expectS[i]))
			c.Check(obtainS[i], check.DeepEquals, expectS[i])
		}
		obtainS = nil
	}
}

func (s *S) TestReadFromFunc(c *check.C) {
	var (
		obtainNfa []string
		obtainSfa [][]alphabet.Letter
	)

	sc := seqio.NewScannerFromFunc(
		fasta.NewReader(
			bytes.NewBufferString(testaln0),
			linear.NewSeq("", nil, alphabet.Protein),
		).Read,
	)
	for sc.Next() {
		t := sc.Seq().(*linear.Seq)
		header := t.Name()
		if desc := t.Description(); len(desc) > 0 {
			header += " " + desc
		}
		obtainNfa = append(obtainNfa, header)
		obtainSfa = append(obtainSfa, t.Slice().(alphabet.Letters))
	}
	c.Check(sc.Error(), check.Equals, nil)
	c.Check(obtainNfa, check.DeepEquals, expectNfa)
	for i := range obtainSfa {
		c.Check(len(obtainSfa[i]), check.Equals, len(expectSfa[i]))
		c.Check(obtainSfa[i], check.DeepEquals, expectSfa[i])
	}
}

func (s *S) TestReadFasta(c *check.C) {
	r := fasta.NewReader(strings.NewReader(fa), linear.NewSeq("", nil, alphabet.Protein))
	m, _ := multi.NewMulti("", nil, seq.DefaultConsensus)
	a, err := NewReader(r, m).Read()
	c.Check(err, check.Equals, nil)
	c.Check(a.Rows(), check.Equals, 11)
}

func main() {
	flag.Parse()
	if *exclude == "" {
		flag.Usage()
		os.Exit(1)
	}

	nameSet := make(map[string]struct{})
	f, err := os.Open(*exclude)
	if err != nil {
		log.Fatalf("failed to open exclude file %q: %v", *exclude, err)
	}
	ls := bufio.NewScanner(f)
	for ls.Scan() {
		nameSet[ls.Text()] = struct{}{}
	}
	err = ls.Err()
	if err != nil {
		log.Fatalf("failed to read exclude file: %v", err)
	}

	sc := seqio.NewScanner(fasta.NewReader(os.Stdin, linear.NewSeq("", nil, alphabet.DNA)))
	for sc.Next() {
		s := sc.Seq().(*linear.Seq)
		if _, ok := nameSet[s.ID]; ok {
			continue
		}
		fmt.Printf("%60a\n", s)
	}
	if err := sc.Error(); err != nil {
		log.Fatalf("error during gff read: %v", err)
	}
}

func ExampleSet_AppendEach() {
	ss := [][]alphabet.Letter{
		[]alphabet.Letter("ACGCTGACTTGGTGCACGT"),
		[]alphabet.Letter("ACGACTGGGACGT"),
		[]alphabet.Letter("ACGCTGACTGGCCGT"),
		[]alphabet.Letter("GCCTTTGCACGT"),
	}
	set = make(Set, 4)
	for i := range set {
		set[i] = linear.NewSeq(fmt.Sprintf("example DNA %d", i), ss[i], alphabet.DNA)
	}
	as := [][]alphabet.QLetter{
		alphabet.QLetter{L: 'A'}.Repeat(2),
		alphabet.QLetter{L: 'C'}.Repeat(2),
		alphabet.QLetter{L: 'G'}.Repeat(2),
		alphabet.QLetter{L: 'T'}.Repeat(2),
	}

	set.AppendEach(as)

	for _, s := range set {
		fmt.Printf("%-s\n", s)
	}
	// Output:
	// ACGCTGACTTGGTGCACGTAA
	// ACGACTGGGACGTCC
	// ACGCTGACTGGCCGTGG
	// GCCTTTGCACGTTT
}

func main() {
	flag.Parse()
	if *in == "" {
		flag.Usage()
		os.Exit(1)
	}

	f, err := os.Open(*in)
	if err != nil {
		log.Fatalf("failed to open %q: %v", *in, err)
	}
	events := make(map[string][]*gff.Feature)
	fsc := featio.NewScanner(gff.NewReader(f))
	for fsc.Next() {
		f := fsc.Feat().(*gff.Feature)
		fields := strings.Fields(f.FeatAttributes.Get("Read"))
		if len(fields) != 3 {
			log.Fatalf("bad record: %+v", f)
		}
		events[fields[0]] = append(events[fields[0]], f)
	}
	if err := fsc.Error(); err != nil {
		log.Fatalf("error during gff read: %v", err)
	}
	f.Close()

	for _, ref := range flag.Args() {
		f, err = os.Open(ref)
		if err != nil {
			log.Fatalf("failed to open reference %q: %v", ref, err)
		}
		ssc := seqio.NewScanner(fasta.NewReader(f, linear.NewSeq("", nil, alphabet.DNA)))
		for ssc.Next() {
			seq := ssc.Seq().(*linear.Seq)
			for _, f := range events[seq.Name()] {
				fields := strings.Fields(f.FeatAttributes.Get("Read"))
				if len(fields) != 3 {
					log.Fatalf("bad record: %+v", f)
				}
				start, err := strconv.Atoi(fields[1])
				if err != nil {
					log.Fatalf("failed to get start coordinate: %v", err)
				}
				end, err := strconv.Atoi(fields[2])
				if err != nil {
					log.Fatalf("failed to get end coordinate: %v", err)
				}
				tmp := *seq
				tmp.ID += fmt.Sprintf("//%d_%d", start, end)
				tmp.Seq = tmp.Seq[start:end]
				fmt.Printf("%60a\n", &tmp)
			}
		}
		if err := ssc.Error(); err != nil {
			log.Fatalf("error during fasta read: %v", err)
		}
		f.Close()
	}
}

func (s *S) SetUpSuite(c *check.C) {
	MaxKmerLen = 14
	s.Seq = linear.NewSeq("", nil, alphabet.DNA)
	s.Seq.Seq = make(alphabet.Letters, testLen)
	for i := range s.Seq.Seq {
		s.Seq.Seq[i] = [...]alphabet.Letter{'A', 'C', 'G', 'T', 'a', 'c', 'g', 't'}[rand.Int()%8]
	}
}

func ExampleNewMulti() {
	m, err := NewMulti("example multi",
		[]seq.Sequence{
			linear.NewSeq("example DNA 1", []alphabet.Letter("ACGCTGACTTGGTGCACGT"), alphabet.DNA),
			linear.NewSeq("example DNA 2", []alphabet.Letter("ACGGTGACCTGGCGCGCAT"), alphabet.DNA),
			linear.NewSeq("example DNA 3", []alphabet.Letter("ACGATGACGTGGCGCTCAT"), alphabet.DNA),
		},
		seq.DefaultConsensus)

	if err != nil {
		return
	}

	fmt.Printf("%- s\n\n%-s\n", m, m.Consensus(false))
	// Output:
	// ACGCTGACTTGGTGCACGT
	// ACGGTGACCTGGCGCGCAT
	// ACGATGACGTGGCGCTCAT
	//
	// acgntgacntggcgcncat
}

func getFasta(fn string) (seq.Sequence, error) {
	fasta_file, err := os.Open(fn)
	if err != nil {
		fmt.Println("Erro ao ler o arquivo", err)
	}
	defer fasta_file.Close()
	var s []alphabet.Letter
	t := linear.NewSeq("", s, alphabet.Protein)
	reader := fasta.NewReader(fasta_file, t)
	seq, _ := reader.Read()
	return seq, nil
}

// writeFlankSeqs writes fasta files containing the sequence of unmapped flanks
// identified in the primary hits provided. cutoff specifies the minimum sequence
// length to consider. left and right specify the filenames for the left and right
// flank fasta sequence files.
func writeFlankSeqs(reads string, hits hitSet, cutoff int, left, right string) error {
	f, err := os.Open(reads)
	if err != nil {
		return err
	}
	defer f.Close()

	lf, err := os.Create(left)
	if err != nil {
		return err
	}
	rf, err := os.Create(right)
	if err != nil {
		return err
	}

	r := fasta.NewReader(f, linear.NewSeq("", nil, alphabet.DNA))
	sc := seqio.NewScanner(r)
	for sc.Next() {
		seq := sc.Seq().(*linear.Seq)
		h, ok := hits[seq.Name()]
		if !ok {
			continue
		}

		all := seq.Seq
		if h.qStart >= cutoff {
			seq.Seq = all[:h.qStart]
			_, err := fmt.Fprintf(lf, "%60a\n", seq)
			if err != nil {
				return err
			}
		}
		if h.qLen-h.qEnd >= cutoff {
			seq.Seq = all[h.qEnd:]
			_, err := fmt.Fprintf(rf, "%60a\n", seq)
			if err != nil {
				return err
			}
		}
	}
	err = sc.Error()
	if err != nil {
		return err
	}
	err = lf.Close()
	if err != nil {
		return err
	}
	return rf.Close()
}

func readFasta(fn string) (name string, seq string, err error) {
	fFasta, err := os.Open(fn)
	defer fFasta.Close()
	if err != nil {
		return "", "", err
	}
	t := linear.NewSeq("", nil, alphabet.Protein)
	reader := fasta.NewReader(fFasta, t)
	s, err := reader.Read()
	if err != nil {
		return "", "", err
	}
	sl := s.(*linear.Seq)
	return sl.Name(), sl.String(), nil
}

func readContigs(file string) (map[string]int, error) {
	f, err := os.Open(file)
	if err != nil {
		return nil, err
	}
	lengths := make(map[string]int)
	sc := seqio.NewScanner(fasta.NewReader(f, linear.NewSeq("", nil, alphabet.DNA)))
	for sc.Next() {
		s := sc.Seq()
		lengths[s.Name()] = s.Len()
	}
	if err != nil {
		log.Fatalf("error during fasta read: %v", err)
	}
	return lengths, nil
}

func (r Row) Clone() seq.Sequence {
	b := make([]alphabet.Letter, r.Len())
	for i, c := range r.Align.Seq {
		b[i] = c[r.Row]
	}
	switch {
	case r.Row < 0:
		panic("under")
	case r.Row >= r.Align.Rows():
		panic("bang over Rows()")
	case r.Row >= len(r.Align.SubAnnotations):

		panic(fmt.Sprintf("bang over len(SubAnns): %d %d", r.Row, len(r.Align.SubAnnotations)))
	}
	return linear.NewSeq(r.Name(), b, r.Alphabet())
}

func readContigs(file string) (map[string]*linear.Seq, error) {
	f, err := os.Open(file)
	if err != nil {
		return nil, err
	}
	seqs := make(map[string]*linear.Seq)
	sc := seqio.NewScanner(fasta.NewReader(f, linear.NewSeq("", nil, alphabet.DNAgapped)))
	for sc.Next() {
		s := sc.Seq().(*linear.Seq)
		seqs[s.ID] = s
	}
	if err != nil {
		return nil, err
	}
	return seqs, nil
}