Files
obitools4/pkg/obiformats/fastseq_write_fastq.go

185 lines
3.9 KiB
Go
Raw Normal View History

2022-01-13 23:27:39 +01:00
package obiformats
import (
"bytes"
"fmt"
"io"
"os"
2023-02-08 13:56:50 +01:00
"sync"
2022-01-13 23:27:39 +01:00
"time"
2022-02-24 12:14:52 +01:00
log "github.com/sirupsen/logrus"
"git.metabarcoding.org/lecasofts/go/obitools/pkg/obiiter"
"git.metabarcoding.org/lecasofts/go/obitools/pkg/obioptions"
2022-01-13 23:43:01 +01:00
"git.metabarcoding.org/lecasofts/go/obitools/pkg/obiseq"
"git.metabarcoding.org/lecasofts/go/obitools/pkg/obiutils"
2022-01-13 23:27:39 +01:00
)
// The function FormatFastq takes a BioSequence object, a quality shift value, and a header formatter
// function as input, and returns a formatted string in FASTQ format.
func FormatFastq(seq *obiseq.BioSequence, formater FormatHeader) string {
2022-01-13 23:27:39 +01:00
q := seq.QualitiesString()
2022-01-13 23:27:39 +01:00
info := ""
if formater != nil {
info = formater(seq)
}
return fmt.Sprintf("@%s %s\n%s\n+\n%s",
2022-01-13 23:27:39 +01:00
seq.Id(), info,
seq.String(),
q,
2022-01-13 23:27:39 +01:00
)
}
func FormatFastqBatch(batch obiiter.BioSequenceBatch, quality_shift int,
formater FormatHeader, skipEmpty bool) []byte {
2022-01-13 23:27:39 +01:00
var bs bytes.Buffer
for _, seq := range batch.Slice() {
if seq.Len() > 0 {
bs.WriteString(FormatFastq(seq, formater))
bs.WriteString("\n")
} else {
if skipEmpty {
log.Warnf("Sequence %s is empty and skiped in output", seq.Id())
} else {
log.Fatalf("Sequence %s is empty", seq.Id())
}
}
2022-01-13 23:27:39 +01:00
}
return bs.Bytes()
}
type FileChunck struct {
text []byte
order int
}
2023-01-22 22:04:17 +01:00
func WriteFastq(iterator obiiter.IBioSequence,
file io.WriteCloser,
2023-01-22 22:04:17 +01:00
options ...WithOption) (obiiter.IBioSequence, error) {
iterator = iterator.Rebatch(1000)
opt := MakeOptions(options)
file, _ = obiutils.CompressStream(file, opt.CompressedFile(), opt.CloseFile())
newIter := obiiter.MakeIBioSequence()
2022-01-13 23:27:39 +01:00
nwriters := opt.ParallelWorkers()
2022-01-13 23:27:39 +01:00
obiiter.RegisterAPipe()
2022-01-13 23:27:39 +01:00
chunkchan := make(chan FileChunck)
header_format := opt.FormatFastSeqHeader()
quality := obioptions.OutputQualityShift()
2022-01-13 23:27:39 +01:00
2022-01-14 17:32:12 +01:00
newIter.Add(nwriters)
2022-01-13 23:27:39 +01:00
2023-02-08 13:56:50 +01:00
var waitWriter sync.WaitGroup
2022-01-13 23:27:39 +01:00
go func() {
newIter.WaitAndClose()
2022-01-13 23:27:39 +01:00
for len(chunkchan) > 0 {
time.Sleep(time.Millisecond)
}
close(chunkchan)
2023-02-08 13:56:50 +01:00
waitWriter.Wait()
2022-01-13 23:27:39 +01:00
}()
2023-01-22 22:04:17 +01:00
ff := func(iterator obiiter.IBioSequence) {
2022-01-13 23:27:39 +01:00
for iterator.Next() {
batch := iterator.Get()
chunk := FileChunck{
FormatFastqBatch(batch, quality, header_format, opt.SkipEmptySequence()),
2022-01-13 23:27:39 +01:00
batch.Order(),
}
chunkchan <- chunk
newIter.Push(batch)
2022-01-13 23:27:39 +01:00
}
2022-01-14 17:32:12 +01:00
newIter.Done()
2022-01-13 23:27:39 +01:00
}
2022-02-24 12:14:52 +01:00
log.Debugln("Start of the fastq file writing")
go ff(iterator)
for i := 0; i < nwriters-1; i++ {
2022-01-13 23:27:39 +01:00
go ff(iterator.Split())
}
next_to_send := 0
received := make(map[int]FileChunck, 100)
2023-02-08 13:56:50 +01:00
waitWriter.Add(1)
2022-01-13 23:27:39 +01:00
go func() {
for chunk := range chunkchan {
if chunk.order == next_to_send {
file.Write(chunk.text)
next_to_send++
chunk, ok := received[next_to_send]
for ok {
file.Write(chunk.text)
delete(received, next_to_send)
next_to_send++
chunk, ok = received[next_to_send]
}
} else {
received[chunk.order] = chunk
}
}
file.Close()
2023-02-08 13:56:50 +01:00
log.Debugln("End of the fastq file writing")
obiiter.UnregisterPipe()
2023-02-08 13:56:50 +01:00
waitWriter.Done()
2022-01-13 23:27:39 +01:00
}()
2022-01-14 17:32:12 +01:00
return newIter, nil
2022-01-13 23:27:39 +01:00
}
2023-01-22 22:04:17 +01:00
func WriteFastqToStdout(iterator obiiter.IBioSequence,
options ...WithOption) (obiiter.IBioSequence, error) {
options = append(options, OptionDontCloseFile())
2022-11-16 17:13:03 +01:00
return WriteFastq(iterator, os.Stdout, options...)
2022-01-13 23:27:39 +01:00
}
2023-01-22 22:04:17 +01:00
func WriteFastqToFile(iterator obiiter.IBioSequence,
2022-01-13 23:27:39 +01:00
filename string,
2023-01-22 22:04:17 +01:00
options ...WithOption) (obiiter.IBioSequence, error) {
2022-01-13 23:27:39 +01:00
opt := MakeOptions(options)
flags := os.O_WRONLY | os.O_CREATE
if opt.AppendFile() {
flags |= os.O_APPEND
}
file, err := os.OpenFile(filename, flags, 0660)
2022-01-13 23:27:39 +01:00
if err != nil {
log.Fatalf("open file error: %v", err)
2023-01-22 22:04:17 +01:00
return obiiter.NilIBioSequence, err
2022-01-13 23:27:39 +01:00
}
options = append(options, OptionCloseFile())
iterator, err = WriteFastq(iterator, file, options...)
if opt.HaveToSavePaired() {
var revfile *os.File
revfile, err = os.OpenFile(opt.PairedFileName(), flags, 0660)
if err != nil {
log.Fatalf("open file error: %v", err)
return obiiter.NilIBioSequence, err
}
iterator, err = WriteFastq(iterator.PairedWith(), revfile, options...)
}
return iterator, err
2022-01-13 23:27:39 +01:00
}