#include "squid/squid.h"
#include "util.h"

Include dependency graph for seq.h:

This graph shows which files directly or indirectly include this file:

Data Structures
struct	mseq_t
	structure for storing multiple sequences More...

Macros
#define	SEQTYPE_UNKNOWN kOtherSeq
#define	SEQTYPE_DNA kDNA
#define	SEQTYPE_RNA kRNA
#define	SEQTYPE_PROTEIN kAmino
#define	AMINOACID_ANY 'X'
#define	NUCLEOTIDE_ANY 'N'

Functions
void	AliStat (mseq_t *prMSeq, bool bSampling, bool bReportAll)
	Stripped down version of squid's alistat.
void	AddSeq (mseq_t *prMSeqDest_p, char pcSeqName, char *pcSeqRes)
	Creates a new sequence entry and appends it to an existing mseq structure.
void	SeqSwap (mseq_t *mseq, int i, int j)
	Swap two sequences in an mseq_t structure.
void	DealignMSeq (mseq_t *mseq)
	Dealigns all sequences in mseq structure, updates the sequence length info and sets aligned to FALSE.
const char *	SeqTypeToStr (int seqtype)
	convert int-encoded iSeqType to string
int	ReadSequences (mseq_t prMSeq_p, char pcSeqFile, int iSeqType, int iSeqFmt, bool bIsProfile, bool bDealignInputSeqs, int iMaxNumSeq, int iMaxSeqLen, char *pcHMMBatch)
	reads sequences from file
void	NewMSeq (mseq_t **mseq)
	allocate and initialise new mseq_t
void	FreeMSeq (mseq_t **mseq)
	Frees an mseq_t and it's members and zeros all members.
void	CopyMSeq (mseq_t *prMSeqDest_p, mseq_t prMSeqSrc)
	copies an mseq structure
void	LogSqInfo (SQINFO *sqinfo)
	debug output of sqinfo struct
int	FindSeqName (char seqname, mseq_t mseq)
int	WriteAlignment (mseq_t mseq, const char aln_outfile, int msafile_format, int iWrap, bool bResno)
	Write alignment to file.
void	DealignSeq (char *seq)
	Removes all gap-characters from a sequence.
void	ShuffleMSeq (mseq_t *prMSeq)
	Shuffle mseq order.
void	SortMSeqByLength (mseq_t *prMSeq, const char cOrder)
	Sort sequences by length.
void	JoinMSeqs (mseq_t *prMSeqDest_p, mseq_t prMSeqToAdd)
	Appends an mseq structure to an already existing one. filename will be left untouched.
bool	SeqsAreAligned (mseq_t *prMSeq, bool bIsProfile, bool bDealignInputSeqs)
	Checks if sequences in given mseq structure are aligned. By definition this is only true, if sequences are of the same length and at least one gap was found.

Macro Definition Documentation

◆ AMINOACID_ANY

#define AMINOACID_ANY 'X'

◆ NUCLEOTIDE_ANY

#define NUCLEOTIDE_ANY 'N'

◆ SEQTYPE_DNA

#define SEQTYPE_DNA kDNA

◆ SEQTYPE_PROTEIN

#define SEQTYPE_PROTEIN kAmino

◆ SEQTYPE_RNA

#define SEQTYPE_RNA kRNA

◆ SEQTYPE_UNKNOWN

#define SEQTYPE_UNKNOWN kOtherSeq

int-encoded sequence types. these are in sync with squid's seqtypes and only used for convenience here

Function Documentation

◆ AddSeq()

void AddSeq	(	mseq_t **	prMSeqDest_p,
		char *	pcSeqName,
		char *	pcSeqRes )

extern

Creates a new sequence entry and appends it to an existing mseq structure.

Parameters

[out]	prMSeqDest_p	Already existing and initialised mseq structure
[in]	pcSeqName	sequence name of the sequence to add
[in]	pcSeqRes	the actual sequence (residues) to add

Note: Don't forget to update the align and type flag if necessary!

FIXME allow adding of more features

◆ AliStat()

void AliStat	(	mseq_t *	prMSeq,
		bool	bSampling,
		bool	bReportAll )

extern

Stripped down version of squid's alistat.

Parameters

[in]	prMSeq	The alignment to analyse
[in]	bSampling	For many sequences: samples from pool
[in]	bReportAll	Report identities for all sequence pairs

Don't have to worry about sequence case because our version of PairwiseIdentity is case insensitive

mseq to squid msa

FIXME code overlap with WriteAlignment. Make it a function and take code there (contains more comments) as template

◆ CopyMSeq()

void CopyMSeq	(	mseq_t **	prMSeqDest_p,
		mseq_t *	prMSeqSrc )

extern

copies an mseq structure

Parameters

[out]	prMSeqDest_p	Copy of mseq structure
[in]	prMSeqSrc	Source mseq structure to copy

Note: caller has to free copy by calling FreeMSeq()

◆ DealignMSeq()

void DealignMSeq ( mseq_t * mseq )

extern

Dealigns all sequences in mseq structure, updates the sequence length info and sets aligned to FALSE.

Parameters

[out] mseq The mseq structure to dealign

◆ DealignSeq()

void DealignSeq ( char * seq )

extern

Removes all gap-characters from a sequence.

Parameters

[out] seq Sequence to dealign

Note: seq will not be reallocated

◆ FindSeqName()

int FindSeqName	(	char *	seqname,
		mseq_t *	mseq )

extern

Parameters

[in]	seqname	The sequence name to search for
[in]	mseq	The multiple sequence structure to search in

Returns: -1 on failure, sequence index of matching name otherwise

Warning: If sequence name happens to be used twice, only the first one will be reported back

◆ FreeMSeq()

void FreeMSeq ( mseq_t ** mseq )

extern

Frees an mseq_t and it's members and zeros all members.

Parameters

[in] mseq mseq_to to free

Note: use in conjunction with NewMSeq()

See also: new_mseq

◆ JoinMSeqs()

void JoinMSeqs	(	mseq_t **	prMSeqDest_p,
		mseq_t *	prMSeqToAdd )

Appends an mseq structure to an already existing one. filename will be left untouched.

Parameters

[in]	prMSeqDest_p	MSeq structure to which to append to
[out]	prMSeqToAdd	MSeq structure which is to append

◆ LogSqInfo()

void LogSqInfo ( SQINFO * sqinfo )

extern

debug output of sqinfo struct

Parameters

[in] sqinfo Squid's SQINFO struct for a certain seqeuence

Note: useful for debugging only

◆ NewMSeq()

void NewMSeq ( mseq_t ** prMSeq )

extern

allocate and initialise new mseq_t

Parameters

[out] prMSeq newly allocated and initialised mseq_t

Note: caller has to free by calling FreeMSeq()

See also: FreeMSeq

◆ ReadSequences()

int ReadSequences	(	mseq_t *	prMSeq,
		char *	seqfile,
		int	iSeqType,
		int	iSeqFmt,
		bool	bIsProfile,
		bool	bDealignInputSeqs,
		int	iMaxNumSeq,
		int	iMaxSeqLen,
		char *	pcHMMBatch )

extern

reads sequences from file

Parameters

[out]	prMSeq	Multiple sequence struct. Must be preallocated. FIXME: would make more sense to allocate it here.
[in]	seqfile	Sequence file name. If '-' sequence will be read from stdin.
[in]	iSeqType	int-encoded sequence type. Set to SEQTYPE_UNKNOWN for autodetect (guessed from first sequence)
[in]	iMaxNumSeq	Return an error, if more than iMaxNumSeq have been read
[in]	iMaxSeqLen	Return an error, if a seq longer than iMaxSeqLen has been read

Returns: 0 on success, -1 on error

Note

Depends heavily on squid
Sequence file format will be guessed
If supported by squid, gzipped files can be read as well.

◆ SeqsAreAligned()

bool SeqsAreAligned	(	mseq_t *	prMSeq,
		bool	bIsProfile,
		bool	bDealignInputSeqs )

Checks if sequences in given mseq structure are aligned. By definition this is only true, if sequences are of the same length and at least one gap was found.

Parameters

[in] prMSeq Sequences to check

Returns: TRUE if sequences are aligned, FALSE if not

◆ SeqSwap()

void SeqSwap	(	mseq_t *	prMSeq,
		int	i,
		int	j )

extern

Swap two sequences in an mseq_t structure.

Parameters

[out]	prMSeq	Multiple sequence struct
[in]	i	Index of first sequence
[in]	j	Index of seconds sequence

◆ SeqTypeToStr()

const char * SeqTypeToStr ( int iSeqType )

extern

convert int-encoded iSeqType to string

Parameters

[in] iSeqType int-encoded sequence type

Returns: character pointer describing the sequence type

◆ ShuffleMSeq()

void ShuffleMSeq ( mseq_t * mseq )

extern

Shuffle mseq order.

Parameters

[out] mseq mseq structure to shuffle

◆ SortMSeqByLength()

void SortMSeqByLength	(	mseq_t *	prMSeq,
		const char	cOrder )

extern

Sort sequences by length.

Parameters

[out]	prMSeq	mseq to sort by length
[out]	cOrder	Sorting order. 'd' for descending, 'a' for ascending.

◆ WriteAlignment()

int WriteAlignment	(	mseq_t *	mseq,
		const char *	pcAlnOutfile,
		int	outfmt,
		int	iWrap,
		bool	bResno )

extern

Write alignment to file.

Parameters

[in]	mseq	The mseq_t struct containing the aligned sequences
[in]	pcAlnOutfile	The name of the output file
[in]	outfmt	The alignment output format (defined in squid.h)
[in]	iWrap	length of line for Clustal/Fasta format

Returns: Non-zero on error

Note: We create a temporary squid MSA struct in here because we never use it within clustal. We might be better of using the old clustal output routines instead.

Data Structures

Macros

Functions

Macro Definition Documentation

◆ AMINOACID_ANY

◆ NUCLEOTIDE_ANY

◆ SEQTYPE_DNA

◆ SEQTYPE_PROTEIN

◆ SEQTYPE_RNA

◆ SEQTYPE_UNKNOWN

Function Documentation

◆ AddSeq()

◆ AliStat()

◆ CopyMSeq()

◆ DealignMSeq()

◆ DealignSeq()

◆ FindSeqName()

◆ FreeMSeq()

◆ JoinMSeqs()

◆ LogSqInfo()

◆ NewMSeq()

◆ ReadSequences()

◆ SeqsAreAligned()

◆ SeqSwap()

◆ SeqTypeToStr()

◆ ShuffleMSeq()

◆ SortMSeqByLength()

◆ WriteAlignment()