src/clustal/seq.c File Reference

#include <assert.h>
#include "squid/squid.h"
#include <ctype.h>
#include "util.h"
#include "log.h"
#include "seq.h"

Include dependency graph for seq.c:

Defines
#define	ALLOW_ONLY_PROTEIN 0
Functions
void	AliStat (mseq_t *prMSeq, bool bSampling, bool bReportAll)
	Stripped down version of squid's alistat.
void	ShuffleMSeq (mseq_t *mseq)
	Shuffle mseq order.
void	SeqSwap (mseq_t *prMSeq, 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.
void	LogSqInfo (SQINFO *sqinfo)
	debug output of sqinfo struct
const char *	SeqTypeToStr (int iSeqType)
	convert int-encoded iSeqType to string
int	ReadSequences (mseq_t prMSeq, char seqfile, int iSeqType, int iSeqFmt, int iMaxNumSeq, int iMaxSeqLen)
	reads sequences from file
void	NewMSeq (mseq_t **prMSeq)
	allocate and initialise new mseq_t
void	CopyMSeq (mseq_t *prMSeqDest_p, mseq_t prMSeqSrc)
	copies an mseq structure
int	FindSeqName (char seqname, mseq_t mseq)
void	FreeMSeq (mseq_t **mseq)
	Frees an mseq_t and it's members and zeros all members.
int	WriteAlignment (mseq_t mseq, const char pcAlnOutfile, int outfmt)
	Write alignment to file.
void	DealignSeq (char *seq)
	Removes all gap-characters from a sequence.
void	SortMSeqByLength (mseq_t *prMSeq, const char cOrder)
	Sort sequences by length.
bool	SeqsAreAligned (mseq_t *prMSeq)
	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.
void	AddSeq (mseq_t *prMSeqDest_p, char pcSeqName, char *pcSeqRes)
	Creates a new sequence entry and appends it to an existing mseq structure.
void	JoinMSeqs (mseq_t *prMSeqDest_p, mseq_t prMSeqToAdd)
	Appends an mseq structure to an already existing one. filename will be left untouched.

Define Documentation

#define ALLOW_ONLY_PROTEIN 0

Function Documentation

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

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

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

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

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

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()

void DealignMSeq ( mseq_t * mseq )

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

Parameters:

[out] mseq The mseq structure to dealign

void DealignSeq ( char * seq )

Removes all gap-characters from a sequence.

Parameters:

[out] seq Sequence to dealign

Note:: seq will not be reallocated

int FindSeqName	(	char *	seqname,
		mseq_t *	mseq
	)

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

void FreeMSeq ( mseq_t ** mseq )

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

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

void LogSqInfo ( SQINFO * sqinfo )

debug output of sqinfo struct

Parameters:

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

Note:: useful for debugging only

void NewMSeq ( mseq_t ** prMSeq )

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

int ReadSequences	(	mseq_t *	prMSeq,
		char *	seqfile,
		int	iSeqType,
		int	iSeqFmt,
		int	iMaxNumSeq,
		int	iMaxSeqLen
	)

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.

bool SeqsAreAligned ( mseq_t * prMSeq )

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

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

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

const char* SeqTypeToStr ( int iSeqType )

convert int-encoded iSeqType to string

Parameters:

[in] iSeqType int-encoded sequence type

Returns:: character pointer describing the sequence type

void ShuffleMSeq ( mseq_t * mseq )

Shuffle mseq order.

Parameters:

[out] mseq mseq structure to shuffle

void SortMSeqByLength	(	mseq_t *	prMSeq,
		const char	cOrder
	)

Sort sequences by length.

Parameters:

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

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

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)

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.

src/clustal/seq.c File Reference

Defines

Functions

Define Documentation

Function Documentation