obitools/obitools4
2
0
Fork 0
mirror of https://github.com/metabarcoding/obitools4.git synced 2025-07-18 17:40:45 +00:00
Code Issues Packages Projects Releases Wiki Activity

Add a reading option on readers to convet U to T

Browse Source
This commit is contained in:
Eric Coissac
2025-07-07 15:29:07 +02:00
parent 8c26fc9884
commit 8d53d253d4
4 changed files with 34 additions and 13 deletions
Download Patch File Download Diff File Expand all files Collapse all files

10
pkg/obiformats/embl_read.go
View File

@@ -87,7 +87,7 @@ func EndOfLastFlatFileEntry(buff []byte) int {
return -1 return -1
} }
func EmblChunkParser(withFeatureTable bool) func(string, io.Reader) (obiseq.BioSequenceSlice, error) { func EmblChunkParser(withFeatureTable, UtoT bool) func(string, io.Reader) (obiseq.BioSequenceSlice, error) {
parser := func(source string, input io.Reader) (obiseq.BioSequenceSlice, error) { parser := func(source string, input io.Reader) (obiseq.BioSequenceSlice, error) {
scanner := bufio.NewScanner(input) scanner := bufio.NewScanner(input)
sequences := make(obiseq.BioSequenceSlice, 0, 100) sequences := make(obiseq.BioSequenceSlice, 0, 100)
@@ -128,6 +128,9 @@ func EmblChunkParser(withFeatureTable bool) func(string, io.Reader) (obiseq.BioS
parts := strings.SplitN(line[5:], " ", 7) parts := strings.SplitN(line[5:], " ", 7)
np := len(parts) - 1 np := len(parts) - 1
for i := 0; i < np; i++ { for i := 0; i < np; i++ {
if UtoT {
parts[i] = strings.ReplaceAll(parts[i], "u", "t")
}
seqBytes.WriteString(parts[i]) seqBytes.WriteString(parts[i])
} }
case line == "//": case line == "//":
@@ -161,10 +164,10 @@ func EmblChunkParser(withFeatureTable bool) func(string, io.Reader) (obiseq.BioS
func _ParseEmblFile( func _ParseEmblFile(
input ChannelFileChunk, input ChannelFileChunk,
out obiiter.IBioSequence, out obiiter.IBioSequence,
withFeatureTable bool, withFeatureTable, UtoT bool,
) { ) {
parser := EmblChunkParser(withFeatureTable) parser := EmblChunkParser(withFeatureTable, UtoT)
for chunks := range input { for chunks := range input {
order := chunks.Order order := chunks.Order
@@ -206,6 +209,7 @@ func ReadEMBL(reader io.Reader, options ...WithOption) (obiiter.IBioSequence, er
entry_channel, entry_channel,
newIter, newIter,
opt.WithFeatureTable(), opt.WithFeatureTable(),
opt.UtoT(),
) )
} }

14
pkg/obiformats/fastaseq_read.go
View File

@@ -39,7 +39,7 @@ func EndOfLastFastaEntry(buffer []byte) int {
return last return last
} }
func FastaChunkParser() func(string, io.Reader) (obiseq.BioSequenceSlice, error) { func FastaChunkParser(UtoT bool) func(string, io.Reader) (obiseq.BioSequenceSlice, error) {
parser := func(source string, input io.Reader) (obiseq.BioSequenceSlice, error) { parser := func(source string, input io.Reader) (obiseq.BioSequenceSlice, error) {
var identifier string var identifier string
@@ -131,7 +131,9 @@ func FastaChunkParser() func(string, io.Reader) (obiseq.BioSequenceSlice, error)
if C >= 'A' && C <= 'Z' { if C >= 'A' && C <= 'Z' {
C = C + 'a' - 'A' C = C + 'a' - 'A'
} }
if UtoT && C == 'u' {
C = 't'
}
if (C >= 'a' && C <= 'z') || C == '-' || C == '.' || C == '[' || C == ']' { if (C >= 'a' && C <= 'z') || C == '-' || C == '.' || C == '[' || C == ']' {
seqBytes.WriteByte(C) seqBytes.WriteByte(C)
} else { } else {
@@ -170,6 +172,9 @@ func FastaChunkParser() func(string, io.Reader) (obiseq.BioSequenceSlice, error)
if C >= 'A' && C <= 'Z' { if C >= 'A' && C <= 'Z' {
C = C + 'a' - 'A' C = C + 'a' - 'A'
} }
if UtoT && C == 'u' {
C = 't'
}
// Removing white space from the sequence // Removing white space from the sequence
if (C >= 'a' && C <= 'z') || C == '-' || C == '.' || C == '[' || C == ']' { if (C >= 'a' && C <= 'z') || C == '-' || C == '.' || C == '[' || C == ']' {
seqBytes.WriteByte(C) seqBytes.WriteByte(C)
@@ -207,9 +212,10 @@ func FastaChunkParser() func(string, io.Reader) (obiseq.BioSequenceSlice, error)
func _ParseFastaFile( func _ParseFastaFile(
input ChannelFileChunk, input ChannelFileChunk,
out obiiter.IBioSequence, out obiiter.IBioSequence,
UtoT bool,
) { ) {
parser := FastaChunkParser() parser := FastaChunkParser(UtoT)
for chunks := range input { for chunks := range input {
sequences, err := parser(chunks.Source, chunks.Raw) sequences, err := parser(chunks.Source, chunks.Raw)
@@ -243,7 +249,7 @@ func ReadFasta(reader io.Reader, options ...WithOption) (obiiter.IBioSequence, e
for i := 0; i < nworker; i++ { for i := 0; i < nworker; i++ {
out.Add(1) out.Add(1)
go _ParseFastaFile(chkchan, out) go _ParseFastaFile(chkchan, out, opt.UtoT())
} }
go func() { go func() {

13
pkg/obiformats/fastqseq_read.go
View File

@@ -131,7 +131,7 @@ func _storeSequenceQuality(bytes *bytes.Buffer, out *obiseq.BioSequence, quality
out.SetQualities(q) out.SetQualities(q)
} }
func FastqChunkParser(quality_shift byte, with_quality bool) func(string, io.Reader) (obiseq.BioSequenceSlice, error) { func FastqChunkParser(quality_shift byte, with_quality bool, UtoT bool) func(string, io.Reader) (obiseq.BioSequenceSlice, error) {
parser := func(source string, input io.Reader) (obiseq.BioSequenceSlice, error) { parser := func(source string, input io.Reader) (obiseq.BioSequenceSlice, error) {
var identifier string var identifier string
@@ -209,6 +209,9 @@ func FastqChunkParser(quality_shift byte, with_quality bool) func(string, io.Rea
if C >= 'A' && C <= 'Z' { if C >= 'A' && C <= 'Z' {
C = C + 'a' - 'A' C = C + 'a' - 'A'
} }
if C == 'u' && UtoT {
C = 't'
}
seqBytes.Reset() seqBytes.Reset()
seqBytes.WriteByte(C) seqBytes.WriteByte(C)
state = 6 state = 6
@@ -228,6 +231,9 @@ func FastqChunkParser(quality_shift byte, with_quality bool) func(string, io.Rea
if C >= 'A' && C <= 'Z' { if C >= 'A' && C <= 'Z' {
C = C + 'a' - 'A' C = C + 'a' - 'A'
} }
if C == 'u' && UtoT {
C = 't'
}
if (C >= 'a' && C <= 'z') || C == '-' || C == '.' || C == '[' || C == ']' { if (C >= 'a' && C <= 'z') || C == '-' || C == '.' || C == '[' || C == ']' {
seqBytes.WriteByte(C) seqBytes.WriteByte(C)
} else { } else {
@@ -301,10 +307,10 @@ func _ParseFastqFile(
input ChannelFileChunk, input ChannelFileChunk,
out obiiter.IBioSequence, out obiiter.IBioSequence,
quality_shift byte, quality_shift byte,
with_quality bool, with_quality, UtoT bool,
) { ) {
parser := FastqChunkParser(quality_shift, with_quality) parser := FastqChunkParser(quality_shift, with_quality, UtoT)
for chunks := range input { for chunks := range input {
sequences, err := parser(chunks.Source, chunks.Raw) sequences, err := parser(chunks.Source, chunks.Raw)
@@ -342,6 +348,7 @@ func ReadFastq(reader io.Reader, options ...WithOption) (obiiter.IBioSequence, e
out, out,
obidefault.ReadQualitiesShift(), obidefault.ReadQualitiesShift(),
opt.ReadQualities(), opt.ReadQualities(),
opt.UtoT(),
) )
} }

10
pkg/obiformats/genbank_read.go
View File

@@ -29,7 +29,7 @@ const (
var _seqlenght_rx = regexp.MustCompile(" +([0-9]+) bp") var _seqlenght_rx = regexp.MustCompile(" +([0-9]+) bp")
func GenbankChunkParser(withFeatureTable bool) func(string, io.Reader) (obiseq.BioSequenceSlice, error) { func GenbankChunkParser(withFeatureTable, UtoT bool) func(string, io.Reader) (obiseq.BioSequenceSlice, error) {
return func(source string, input io.Reader) (obiseq.BioSequenceSlice, error) { return func(source string, input io.Reader) (obiseq.BioSequenceSlice, error) {
state := inHeader state := inHeader
scanner := bufio.NewReader(input) scanner := bufio.NewReader(input)
@@ -165,6 +165,9 @@ func GenbankChunkParser(withFeatureTable bool) func(string, io.Reader) (obiseq.B
parts := strings.SplitN(line[10:], " ", 6) parts := strings.SplitN(line[10:], " ", 6)
lparts := len(parts) lparts := len(parts)
for i := 0; i < lparts; i++ { for i := 0; i < lparts; i++ {
if UtoT {
parts[i] = strings.ReplaceAll(parts[i], "u", "t")
}
seqBytes.WriteString(parts[i]) seqBytes.WriteString(parts[i])
} }
processed = true processed = true
@@ -200,9 +203,9 @@ func GenbankChunkParser(withFeatureTable bool) func(string, io.Reader) (obiseq.B
func _ParseGenbankFile(input ChannelFileChunk, func _ParseGenbankFile(input ChannelFileChunk,
out obiiter.IBioSequence, out obiiter.IBioSequence,
withFeatureTable bool) { withFeatureTable, UtoT bool) {
parser := GenbankChunkParser(withFeatureTable) parser := GenbankChunkParser(withFeatureTable, UtoT)
for chunks := range input { for chunks := range input {
sequences, err := parser(chunks.Source, chunks.Raw) sequences, err := parser(chunks.Source, chunks.Raw)
@@ -242,6 +245,7 @@ func ReadGenbank(reader io.Reader, options ...WithOption) (obiiter.IBioSequence,
entry_channel, entry_channel,
newIter, newIter,
opt.WithFeatureTable(), opt.WithFeatureTable(),
opt.UtoT(),
) )
} }