Clusters a set of Y-STR samples. More...

#include <config.h>
#include <stdlib.h>
#include <stdio.h>
#include <assert.h>
#include <inttypes.h>
#include <string.h>
#include <math.h>
#include <libxml/tree.h>
#include <jwsc/base/error.h>
#include <jwsc/base/option.h>
#include <jwsc/base/limits.h>
#include <jwsc/vector/vector.h>
#include <jwsc/matrix/matrix.h>
#include <jwsc/matrix/matrix_io.h>
#include <jwsc/matblock/matblock.h>
#include <jwsc/stat/kmeans.h>
#include <jwsc/stat/gmm.h>
#include <jwsc/stat/mmm.h>
#include "haplo_groups.h"
#include "options.h"
#include "output.h"
#include "input.h"
#include "xml.h"

Defines
#define	NUM_OPTS_NO_ARG 0 + NUM_SHARED_OPTS_NO_ARG
#define	NUM_OPTS_WITH_ARG 12 + NUM_SHARED_OPTS_WITH_ARG
#define	LABEL_COL 0
#define	NUM_CLUSTERS 4
#define	CLUSTER_TYPE HAPLO_CLUSTER_KMEANS
#define	MEANS_OUT_FNAME "/dev/null"
#define	WEIGHTS_OUT_FNAME "/dev/null"
#define	RESPONSES_OUT_FNAME "/dev/null"
#define	MEMBERS_OUT_FNAME "/dev/stdout"
Enumerations
enum	Haplo_cluster_type { HAPLO_CLUSTER_KMEANS, HAPLO_CLUSTER_GMM, HAPLO_CLUSTER_MMM }
	Types of clustering supported. More...
Functions
uint32_t	get_num_opts_no_arg ()
	Returns the number of options not taking an argument.
uint32_t	get_num_opts_with_arg ()
	Returns the number of options taking an argument.
void	print_usage ()
	Prints the program usage to stderr.
static Error *	process_num_clusters_opt (Option_arg arg)
	Process the 'num-clusters' program option.
static Error *	process_cluster_type_opt (Option_arg arg)
	Process the 'cluster-type' program option.
Error *	process_means_out_opt (Option_arg arg)
	Process the 'means-out' program option.
Error *	process_weights_out_opt (Option_arg arg)
	Process the 'weights-out' program option.
Error *	process_responses_out_opt (Option_arg arg)
	Process the 'responses-out' program option.
Error *	process_members_out_opt (Option_arg arg)
	Process the 'members-out' program option.
static void	init_cluster_options ()
	Initializes the haplo-cluster program options.
static void	write_means (const Matrix_d *means)
	Writes the clustering means.
static void	write_mmm_means (const Matblock_d *means)
	Writes the MMM clustering means.
static void	write_members_header (const Matblock_u8 ids, const Vector_u32 labels, FILE *fp)
	Writes the file header information for kmeans membership.
static void	write_members (const Matblock_u8 ids, const Vector_u32 labels, const Vector_i32 *members)
	Writes the clustering membership.
static void	write_weights (const Vector_d *weights)
	Writes the clustering weights.
static void	write_responses_header (const Matblock_u8 ids, const Vector_u32 labels, const Matrix_d responses, FILE fp)
	Write the file header for cluster responsibilities.
static void	write_responses (const Matblock_u8 ids, const Vector_u32 labels, const Matrix_d *responses)
	Writes the clustering responsibilities.
static void	write_kmeans_results (const Matblock_u8 ids, const Vector_u32 labels, const Matrix_d means, const Vector_i32 members)
	Writes the kmeans results to files.
static void	assign_gmm_members (Vector_i32 *members_out, const Matrix_d responses)
	Assigns hard cluster membership from GMM responsibilities.
static void	assign_mmm_members (Vector_i32 *members_out, const Matrix_d responses)
	Assigns hard cluster membership from MMM responsibilities.
static void	assign_mmm_aux_members (Vector_i32 *members_out, const Matblock_d means, const Vector_d weights, const Matblock_u32 markers)
	Assigns hard cluster membership from MMM responsibilities.
static void	write_gmm_results (const Matblock_u8 ids, const Vector_u32 labels, const Matrix_d means, const Matblock_d covars, const Vector_d weights, const Matrix_d responses, const Vector_i32 *members)
	Writes the Gaussian mixture model results to files.
static void	write_mmm_results (const Matblock_u8 ids, const Vector_u32 labels, const Matblock_d means, const Vector_d weights, const Matrix_d responses, const Vector_i32 members)
	Writes the multinomial mixture model results to files.
static void	write_mmm_aux_results (const Matblock_u8 ids, const Vector_u32 labels, const Vector_i32 *members)
	Writes the auxilliary multinomial mixture model results to files.
static void	cluster_kmeans (const Matblock_u8 ids, const Vector_u32 labels, const Matrix_i32 *markers)
static void	cluster_gmm (const Matblock_u8 ids, const Vector_u32 labels, const Matrix_i32 *markers)
static void	cluster_mmm (const Matblock_u8 ids, const Vector_u32 labels, const Matrix_i32 markers, const Matblock_u8 aux_ids, const Vector_u32 aux_labels, const Matrix_i32 aux_markers)
int	main (int argc, const char **argv)
	Main function for the haplo-cluster program.
Variables
Option_no_arg	opts_no_arg [NUM_OPTS_NO_ARG]
	Program options not taking an argument.
Option_with_arg	opts_with_arg [NUM_OPTS_WITH_ARG]
	Program options taking an argument.
static uint32_t	num_clusters = NUM_CLUSTERS
	Number of clusters to use in the K-means algorithm.
static Haplo_cluster_type	cluster_type = CLUSTER_TYPE
	Type of clustering to use.
static const char *	means_out_fname = MEANS_OUT_FNAME
	Cluster means output file name.
static const char *	weights_out_fname = WEIGHTS_OUT_FNAME
	Cluster weights output file name.
static const char *	responses_out_fname = RESPONSES_OUT_FNAME
	Cluster responses output file name.
static const char *	members_out_fname = MEMBERS_OUT_FNAME
	Cluster members output file name.

Detailed Description

Clusters a set of Y-STR samples.

Author:: Joseph Schlecht

License:: Creative Commons BY-NC-SA 3.0

Available algorithms for clustering include K-means, Gaussian mixture model, and multinomial mixture model.

Definition in file haplo_cluster.c.

Define Documentation

#define NUM_OPTS_NO_ARG 0 + NUM_SHARED_OPTS_NO_ARG

Definition at line 82 of file haplo_cluster.c.

#define NUM_OPTS_WITH_ARG 12 + NUM_SHARED_OPTS_WITH_ARG

Definition at line 83 of file haplo_cluster.c.

#define LABEL_COL 0

Definition at line 86 of file haplo_cluster.c.

#define NUM_CLUSTERS 4

Definition at line 87 of file haplo_cluster.c.

#define CLUSTER_TYPE HAPLO_CLUSTER_KMEANS

Definition at line 88 of file haplo_cluster.c.

#define MEANS_OUT_FNAME "/dev/null"

Definition at line 89 of file haplo_cluster.c.

#define WEIGHTS_OUT_FNAME "/dev/null"

Definition at line 90 of file haplo_cluster.c.

#define RESPONSES_OUT_FNAME "/dev/null"

Definition at line 91 of file haplo_cluster.c.

#define MEMBERS_OUT_FNAME "/dev/stdout"

Definition at line 92 of file haplo_cluster.c.

Enumeration Type Documentation

enum Haplo_cluster_type

Types of clustering supported.

Enumerator:

HAPLO_CLUSTER_KMEANS
HAPLO_CLUSTER_GMM
HAPLO_CLUSTER_MMM

Definition at line 96 of file haplo_cluster.c.

Function Documentation

uint32_t get_num_opts_no_arg ( )

Returns the number of options not taking an argument.

Definition at line 131 of file haplo_cluster.c.

uint32_t get_num_opts_with_arg ( )

Returns the number of options taking an argument.

Definition at line 137 of file haplo_cluster.c.

void print_usage ( void )

Prints the program usage to stderr.

Definition at line 143 of file haplo_cluster.c.

static Error* process_num_clusters_opt ( Option_arg arg ) [static]

Process the 'num-clusters' program option.

Definition at line 151 of file haplo_cluster.c.

static Error* process_cluster_type_opt ( Option_arg arg ) [static]

Process the 'cluster-type' program option.

Definition at line 165 of file haplo_cluster.c.

Error* process_means_out_opt ( Option_arg arg )

Process the 'means-out' program option.

Definition at line 191 of file haplo_cluster.c.

Error* process_weights_out_opt ( Option_arg arg )

Process the 'weights-out' program option.

Definition at line 202 of file haplo_cluster.c.

Error* process_responses_out_opt ( Option_arg arg )

Process the 'responses-out' program option.

Definition at line 213 of file haplo_cluster.c.

Error* process_members_out_opt ( Option_arg arg )

Process the 'members-out' program option.

Definition at line 224 of file haplo_cluster.c.

static void init_cluster_options ( ) [static]

Initializes the haplo-cluster program options.

Definition at line 235 of file haplo_cluster.c.

static void write_means ( const Matrix_d * means ) [static]

Writes the clustering means.

Definition at line 325 of file haplo_cluster.c.

static void write_mmm_means ( const Matblock_d * means ) [static]

Writes the MMM clustering means.

Definition at line 389 of file haplo_cluster.c.

static void write_members_header	(	const Matblock_u8 *	ids,
		const Vector_u32 *	labels,
		FILE *	fp
	)		`[static]`

Writes the file header information for kmeans membership.

Definition at line 469 of file haplo_cluster.c.

static void write_members	(	const Matblock_u8 *	ids,
		const Vector_u32 *	labels,
		const Vector_i32 *	members
	)		`[static]`

Writes the clustering membership.

Definition at line 528 of file haplo_cluster.c.

static void write_weights ( const Vector_d * weights ) [static]

Writes the clustering weights.

Definition at line 596 of file haplo_cluster.c.

static void write_responses_header	(	const Matblock_u8 *	ids,
		const Vector_u32 *	labels,
		const Matrix_d *	responses,
		FILE *	fp
	)		`[static]`

Write the file header for cluster responsibilities.

Definition at line 644 of file haplo_cluster.c.

static void write_responses	(	const Matblock_u8 *	ids,
		const Vector_u32 *	labels,
		const Matrix_d *	responses
	)		`[static]`

Writes the clustering responsibilities.

Definition at line 713 of file haplo_cluster.c.

static void write_kmeans_results	(	const Matblock_u8 *	ids,
		const Vector_u32 *	labels,
		const Matrix_d *	means,
		const Vector_i32 *	members
	)		`[static]`

Writes the kmeans results to files.

Definition at line 798 of file haplo_cluster.c.

static void assign_gmm_members	(	Vector_i32 **	members_out,
		const Matrix_d *	responses
	)		`[static]`

Assigns hard cluster membership from GMM responsibilities.

Definition at line 811 of file haplo_cluster.c.

static void assign_mmm_members	(	Vector_i32 **	members_out,
		const Matrix_d *	responses
	)		`[static]`

Assigns hard cluster membership from MMM responsibilities.

Definition at line 837 of file haplo_cluster.c.

static void assign_mmm_aux_members	(	Vector_i32 **	members_out,
		const Matblock_d *	means,
		const Vector_d *	weights,
		const Matblock_u32 *	markers
	)		`[static]`

Assigns hard cluster membership from MMM responsibilities.

Definition at line 863 of file haplo_cluster.c.

static void write_gmm_results	(	const Matblock_u8 *	ids,
		const Vector_u32 *	labels,
		const Matrix_d *	means,
		const Matblock_d *	covars,
		const Vector_d *	weights,
		const Matrix_d *	responses,
		const Vector_i32 *	members
	)		`[static]`

Writes the Gaussian mixture model results to files.

Definition at line 927 of file haplo_cluster.c.

static void write_mmm_results	(	const Matblock_u8 *	ids,
		const Vector_u32 *	labels,
		const Matblock_d *	means,
		const Vector_d *	weights,
		const Matrix_d *	responses,
		const Vector_i32 *	members
	)		`[static]`

Writes the multinomial mixture model results to files.

Definition at line 945 of file haplo_cluster.c.

static void write_mmm_aux_results	(	const Matblock_u8 *	ids,
		const Vector_u32 *	labels,
		const Vector_i32 *	members
	)		`[static]`

Writes the auxilliary multinomial mixture model results to files.

Definition at line 962 of file haplo_cluster.c.

static void cluster_kmeans	(	const Matblock_u8 *	ids,
		const Vector_u32 *	labels,
		const Matrix_i32 *	markers
	)		`[static]`

Definition at line 972 of file haplo_cluster.c.

static void cluster_gmm	(	const Matblock_u8 *	ids,
		const Vector_u32 *	labels,
		const Matrix_i32 *	markers
	)		`[static]`

Definition at line 1006 of file haplo_cluster.c.

static void cluster_mmm	(	const Matblock_u8 *	ids,
		const Vector_u32 *	labels,
		const Matrix_i32 *	markers,
		const Matblock_u8 *	aux_ids,
		const Vector_u32 *	aux_labels,
		const Matrix_i32 *	aux_markers
	)		`[static]`

Definition at line 1047 of file haplo_cluster.c.

int main	(	int	argc,
		const char **	argv
	)

Main function for the haplo-cluster program.

Definition at line 1140 of file haplo_cluster.c.

Variable Documentation

Option_no_arg opts_no_arg[NUM_OPTS_NO_ARG]

Program options not taking an argument.

Definition at line 106 of file haplo_cluster.c.

Option_with_arg opts_with_arg[NUM_OPTS_WITH_ARG]

Program options taking an argument.

Definition at line 109 of file haplo_cluster.c.

uint32_t num_clusters = NUM_CLUSTERS [static]

Number of clusters to use in the K-means algorithm.

Definition at line 112 of file haplo_cluster.c.

Haplo_cluster_type cluster_type = CLUSTER_TYPE [static]

Type of clustering to use.

Definition at line 115 of file haplo_cluster.c.

const char* means_out_fname = MEANS_OUT_FNAME [static]

Cluster means output file name.

Definition at line 118 of file haplo_cluster.c.

const char* weights_out_fname = WEIGHTS_OUT_FNAME [static]

Cluster weights output file name.

Definition at line 121 of file haplo_cluster.c.

const char* responses_out_fname = RESPONSES_OUT_FNAME [static]

Cluster responses output file name.

Definition at line 124 of file haplo_cluster.c.

const char* members_out_fname = MEMBERS_OUT_FNAME [static]

Cluster members output file name.

Definition at line 127 of file haplo_cluster.c.

Defines

Enumerations

Functions

Variables

Detailed Description

Define Documentation

Enumeration Type Documentation

Function Documentation

Variable Documentation