func (s *S) TestAlignment(c *check.C) {
l := [...]byte{'A', 'C', 'G', 'T'}
Q := len(l)
a := &linear.Seq{Seq: make(alphabet.Letters, 0, util.Pow(Q, k))}
a.Alpha = alphabet.DNA
for _, i := range util.DeBruijn(byte(Q), k) {
a.Seq = append(a.Seq, alphabet.Letter(l[i]))
}
b := &linear.Seq{Seq: make(alphabet.Letters, 0, util.Pow(Q, k-1))}
b.Alpha = alphabet.DNA
for _, i := range util.DeBruijn(byte(Q), k-1) {
b.Seq = append(b.Seq, alphabet.Letter(l[i]))
}
aligner := NewAligner(a, b, int(k), 50, 0.80)
aligner.Costs = &Costs{
MaxIGap: maxIGap,
DiffCost: diffCost,
SameCost: sameCost,
MatchCost: matchCost,
BlockCost: blockCost,
RMatchCost: rMatchCost,
}
hits := aligner.AlignTraps(T)
c.Check(hits, check.DeepEquals, H)
la, lb, err := hits.Sum()
c.Check(la, check.Equals, 791)
c.Check(lb, check.Equals, 664)
c.Check(err, check.Equals, nil)
for _, h := range H {
sa, sb := &linear.Seq{Seq: a.Seq[h.Abpos:h.Aepos]}, &linear.Seq{Seq: b.Seq[h.Bbpos:h.Bepos]}
sa.Alpha = alphabet.DNAgapped
sb.Alpha = alphabet.DNAgapped
smith := align.SW{
{0, -1, -1, -1, -1},
{-1, 2, -1, -1, -1},
{-1, -1, 2, -1, -1},
{-1, -1, -1, 2, -1},
{-1, -1, -1, -1, 2},
}
swa, _ := smith.Align(sa, sb)
fa := align.Format(sa, sb, swa, sa.Alpha.Gap())
c.Logf("%v\n", swa)
c.Logf("%s\n%s\n", fa[0], fa[1])
}
}
// Finalise the sequence packing.
func (pa *Packer) FinalisePack() *Packed {
lastPad := 0
seq := make(alphabet.Letters, 0, pa.length)
for _, c := range pa.packed.seqMap.contigs {
padding := binSize - c.Len()%binSize
if padding < minPadding {
padding += binSize
}
seq = append(seq, alphabet.Letter('N').Repeat(lastPad)...)
seq = append(seq, c.Seq.Seq...)
lastPad = padding
}
pa.packed.Seq.Seq = seq
return pa.packed
}
func (s *S) TestPack(c *check.C) {
p := NewPacker("")
for k := byte(1); k <= maxk; k++ {
a := &linear.Seq{
Annotation: seq.Annotation{
ID: fmt.Sprintf("deBruijn%d", k),
},
Seq: make(alphabet.Letters, 0, util.Pow(Q, k)),
}
for _, i := range util.DeBruijn(byte(Q), k) {
a.Seq = append(a.Seq, alphabet.Letter(l[i]))
}
ps, _ := p.Pack(a)
c.Logf("%d: %s", k, ps)
}
c.Check(p.FinalisePack().Len(), check.Equals, 94208)
}
func (s *S) SetUpSuite(c *check.C) {
p := NewPacker("")
for k := byte(1); k <= maxk; k++ {
a := &linear.Seq{
Annotation: seq.Annotation{
ID: fmt.Sprintf("deBruijn%d", k),
},
Seq: make(alphabet.Letters, 0, util.Pow(Q, k)),
}
for _, i := range util.DeBruijn(byte(Q), k) {
a.Seq = append(a.Seq, alphabet.Letter(l[i]))
}
a.Loc = Contig(a.ID)
p.Pack(a)
}
ps = p.FinalisePack()
}
func (s *S) TestFilterAndMerge(c *check.C) {
l := [...]byte{'A', 'C', 'G', 'T'}
Q := len(l)
a := &linear.Seq{Seq: make(alphabet.Letters, 0, util.Pow(Q, k))}
a.Alpha = alphabet.DNA
for _, i := range util.DeBruijn(byte(Q), k) {
a.Seq = append(a.Seq, alphabet.Letter(l[i]))
}
b := &linear.Seq{Seq: make(alphabet.Letters, 0, util.Pow(Q, k-1))}
// b.Alpha = alphabet.DNA // Not actually required for this use.
for _, i := range util.DeBruijn(byte(Q), k-1) {
b.Seq = append(b.Seq, alphabet.Letter(l[i]))
}
i, err := kmerindex.New(int(k), a)
if err != nil {
c.Fatalf("Failed to create kmerindex: %v", err)
}
i.Build()
p := &Params{WordSize: int(k), MinMatch: 50, MaxError: 4, TubeOffset: 32}
f := New(i, p)
var sorter *morass.Morass
if sorter, err = morass.New(Hit{}, "", "", 2<<20, false); err != nil {
c.Fatalf("Failed to create morass: %v", err)
}
f.Filter(b, false, false, sorter)
c.Check(sorter.Len(), check.Equals, int64(12))
r := make([]Hit, 1, sorter.Len())
for {
err = sorter.Pull(&r[len(r)-1])
if err != nil {
r = r[:len(r)-1]
break
}
r = append(r, Hit{})
}
want := map[Hit]bool{
Hit{From: 0, To: 163, Diagonal: 32}: true,
Hit{From: 141, To: 247, Diagonal: 64}: true,
Hit{From: 237, To: 433, Diagonal: 1120}: true,
Hit{From: 241, To: 347, Diagonal: 96}: true,
Hit{From: 341, To: 452, Diagonal: 128}: true,
Hit{From: 447, To: 565, Diagonal: 1952}: true,
Hit{From: 542, To: 628, Diagonal: 1984}: true,
Hit{From: 627, To: 814, Diagonal: 2592}: true,
Hit{From: 786, To: 898, Diagonal: 2624}: true,
Hit{From: 868, To: 939, Diagonal: 2880}: true,
Hit{From: 938, To: 997, Diagonal: 3040}: true,
Hit{From: 938, To: 1024, Diagonal: 3072}: true,
}
got := make(map[Hit]bool)
for _, h := range r {
got[h] = true
}
c.Check(got, check.DeepEquals, want)
c.Check(sort.IsSorted(hits(r)), check.Equals, true)
m := NewMerger(i, b, p, 5, false)
for _, h := range r {
m.MergeFilterHit(&h)
}
t := m.FinaliseMerge()
sorter.CleanUp()
c.Check(len(t), check.Equals, 6)
la, lb := t.Sum()
c.Check(la, check.Equals, 1257)
c.Check(lb, check.Equals, 402)
c.Check(t, check.DeepEquals, Trapezoids{
{Top: 452, Bottom: 0, Left: -128, Right: 3},
{Top: 433, Bottom: 237, Left: -1120, Right: -1085},
{Top: 628, Bottom: 447, Left: -1984, Right: -1917},
{Top: 898, Bottom: 627, Left: -2624, Right: -2557},
{Top: 939, Bottom: 868, Left: -2880, Right: -2845},
{Top: 1024, Bottom: 938, Left: -3072, Right: -3005},
})
}
// Read a single sequence and return it and potentially an error. Note that
// a non-nil returned error may be associated with a valid sequence, so it is
// the responsibility of the caller to examine the error to determine whether
// the read was successful.
// Note that if the Reader's template type returns different non-nil error
// values from calls to SetName and SetDescription, a new error string will be
// returned on each call to Read. So to allow direct error comparison these
// methods should return the same error.
// TODO: Does not read multi-line fastq.
func (r *Reader) Read() (seq.Sequence, error) {
const (
id1 = iota
letters
id2
quality
)
var (
buff, line, label []byte
isPrefix bool
seqBuff []alphabet.QLetter
t seqio.SequenceAppender
state int
err error
)
loop:
for {
buff, isPrefix, err = r.r.ReadLine()
if err != nil {
if t != nil && state == quality && err == io.EOF {
err = nil
break
}
return nil, err
}
line = append(line, buff...)
if isPrefix {
continue
}
line = bytes.TrimSpace(line)
switch {
case state == id1 && maybeID1(line):
state = letters
var _err error
t, _err = r.readHeader(line)
if err == nil && _err != nil {
err = _err
}
label = append([]byte(nil), line...)
case state == id2 && maybeID2(line):
state = quality
if len(label) == 0 {
return nil, errors.New("fastq: no header line parsed before +line in fastq format")
}
if len(line) != 1 && bytes.Compare(label[1:], line[1:]) != 0 {
return nil, errors.New("fastq: quality header does not match sequence header")
}
case state == letters && len(line) > 0:
if maybeID2(line) && (len(line) == 1 || bytes.Compare(label[1:], line[1:]) == 0) {
state = quality
break
}
state = id2
seqBuff = make([]alphabet.QLetter, len(line))
var i int
for _, l := range line {
if isSpace(l) {
continue
}
seqBuff[i].L = alphabet.Letter(l)
i++
}
seqBuff = seqBuff[:i]
case state == quality:
if len(line) == 0 && len(seqBuff) != 0 {
continue
}
break loop
}
line = line[:0]
}
line = bytes.Join(bytes.Fields(line), nil)
if len(line) != len(seqBuff) {
return nil, errors.New("fastq: sequence/quality length mismatch")
}
for i := range line {
seqBuff[i].Q = r.enc.DecodeToQphred(line[i])
}
t.AppendQLetters(seqBuff...)
return t, err
}