void vcf_read_header(gzFile vcf_fh, VCFInfo *vcf_info) {
char *line, *cur, *token;
int tok_num;
int n_fix_header, i;
/* const char delim[] = " \t"; */
const char delim[] = "\t";
n_fix_header = sizeof(vcf_fix_headers) / sizeof(const char *);
vcf_info->n_header_line = 0;
while(util_gzgetline(vcf_fh, &vcf_info->buf, &vcf_info->buf_size) != -1) {
line = vcf_info->buf;
if(util_str_starts_with(line, "##")) {
/* header line */
vcf_info->n_header_line += 1;
}
else if(util_str_starts_with(line, "#CHROM")) {
/* this should be last header line that contains list of fixed fields */
vcf_info->n_header_line += 1;
cur = vcf_info->buf;
line = util_str_dup(vcf_info->buf);
tok_num = 0;
while((token = strsep(&cur, delim)) != NULL) {
if(tok_num < n_fix_header) {
if(strcmp(token, vcf_fix_headers[tok_num]) != 0) {
my_warn("expected token %d to be %s but got '%s'",
tok_num, vcf_fix_headers[tok_num], token);
}
}
tok_num += 1;
}
vcf_info->n_sample = tok_num - n_fix_header;
/*
* read sample names from remaining part of header
*/
vcf_info->sample_names = my_malloc(sizeof(char *) * vcf_info->n_sample);
cur = line;
tok_num = 0;
i = 0;
while((token = strsep(&cur, delim)) != NULL) {
if(tok_num >= n_fix_header) {
vcf_info->sample_names[i] = util_str_dup(token);
i += 1;
}
tok_num += 1;
}
my_free(line);
break;
} else {
my_err("expected last line in header to start with #CHROM");
}
}
}
/**
* Gets next line of IMPUTE file and parses it into ImputeInfo datastructure.
*
* IMPUTE files are described here:
* http://www.stats.ox.ac.uk/~marchini/software/gwas/file_format.html
*
* example line:
* --- rs149201999 16050408 T C 0.966 0.034 0 0.395 0.467 ....
*
* If geno_probs array is non-null genotype probabilities are parsed and
* stored in the provided array. The array must be of length
* n_samples*3.
*
* If haplotypes array is non-null phased genotypes are parsed and
* stored in the provided array. The array must be of length
* n_samples*2.
*
* IMPUTE files contain EITHER haplotypes OR genotypes so only
* one of geno_probs or haplotypes should be non-null (at most).
*
* Returns 0 on success, -1 if at EOF.
*/
int impute_read_line(gzFile fh, ImputeInfo *impute_info, SNP *snp,
float *geno_probs, char *haplotypes) {
char *cur, *token;
int n_fix_header, ref_len, alt_len;
size_t tok_num;
const char delim[] = " \t";
/* read a line */
if(util_gzgetline(fh, &impute_info->buf, &impute_info->buf_size) == -1) {
return -1;
}
cur = impute_info->buf;
tok_num = 0;
/* SNP name, often just set to "---" */
token = strsep(&cur, delim);
if(token == NULL) {
my_err("expected at least %d tokens per line\n", IMPUTE_FIX_HEADER);
}
/* SNP identifier (rs_id) */
token = strsep(&cur, delim);
if(token == NULL) {
my_err("expected at least %d tokens per line\n", IMPUTE_FIX_HEADER);
}
util_strncpy(snp->name, token, sizeof(snp->name));
/* pos */
token = strsep(&cur, delim);
if(token == NULL) {
my_err("expected at least %d tokens per line\n", IMPUTE_FIX_HEADER);
}
snp->pos = util_parse_long(token);
/* ref allele */
token = strsep(&cur, delim);
if(token == NULL) {
my_err("expected at least %d tokens per line\n", IMPUTE_FIX_HEADER);
}
util_strncpy(snp->allele1, token, sizeof(snp->allele1));
/* alt allele */
token = strsep(&cur, delim);
if(token == NULL) {
my_err("expected at least %d tokens per line\n", IMPUTE_FIX_HEADER);
}
alt_len = util_strncpy(snp->allele2, token, sizeof(snp->allele2));
/* now parse haplotypes and/or genotype likelihoods */
if(geno_probs && haplotypes) {
my_err("impute2 files contain EITHER genotypes or haplotypes, but "
"both requested\n");
}
else if(geno_probs) {
impute_parse_geno_probs(geno_probs, cur, impute_info->n_samples);
}
else if(haplotypes) {
impute_parse_haplotypes(haplotypes, cur, impute_info->n_samples);
}
}
/**
* Gets next line of VCF file and parses it into VCFInfo datastructure.
*
* If geno_probs array is non-null genotype likelihoods are parsed and
* stored in the provided array. The array must be of length
* n_sample*3.
*
* If haplotypes array is non-null phased genotypes are parsed and
* stored in the provided array. The array must be of length
* n_sample*2.
*
* Returns 0 on success, -1 if at EOF.
*/
int vcf_read_line(gzFile vcf_fh, VCFInfo *vcf_info, SNP *snp,
float *geno_probs, char *haplotypes) {
char *cur, *token;
int n_fix_header, ref_len, alt_len;
size_t tok_num;
/* Used to allow space or tab delimiters here but now only allow
* tab. This is because VCF specification indicates that fields
* should be tab-delimited, and occasionally some fields contain
* spaces.
*/
/* const char delim[] = " \t";*/
const char delim[] = "\t";
n_fix_header = sizeof(vcf_fix_headers) / sizeof(const char *);
/* read a line */
if(util_gzgetline(vcf_fh, &vcf_info->buf, &vcf_info->buf_size) == -1) {
return -1;
}
cur = vcf_info->buf;
tok_num = 0;
/* chrom */
token = strsep(&cur, delim);
if(token == NULL) {
my_err("expected at least %d tokens per line\n", n_fix_header);
}
/* we don't bother to store chromosome since we store
* SNPs from each chromosome in their own table
*/
/* util_strncpy(snp->chrom, token, sizeof(snp->chrom)); */
/* pos */
token = strsep(&cur, delim);
if(token == NULL) {
my_err("expected at least %d tokens per line\n", n_fix_header);
}
snp->pos = util_parse_long(token);
/* ID */
token = strsep(&cur, delim);
if(token == NULL) {
my_err("expected at least %d tokens per line\n", n_fix_header);
}
util_strncpy(snp->name, token, sizeof(snp->name));
/* ref */
token = strsep(&cur, delim);
if(token == NULL) {
my_err("expected at least %d tokens per line\n", n_fix_header);
}
ref_len = util_strncpy(snp->allele1, token, sizeof(snp->allele1));
/* used to warn about truncations, but makes program too
* chatty if there are a lot of them
*/
vcf_info->ref_len = 0;
/* vcf_info->ref_len = strlen(token); */
/* if(ref_len != vcf_info->ref_len) { */
/* my_warn("truncating long allele (%ld bp) to %ld bp\n", */
/* vcf_info->ref_len, ref_len); */
/* } */
/* alt */
token = strsep(&cur, delim);
if(token == NULL) {
my_err("expected at least %d tokens per line\n", n_fix_header);
}
alt_len = util_strncpy(snp->allele2, token, sizeof(snp->allele2));
vcf_info->alt_len = 0;
/* vcf_info->alt_len = strlen(token); */
/* if(alt_len != vcf_info->alt_len) { */
/* my_warn("truncating long allele (%ld bp) to %ld bp\n", */
/* vcf_info->alt_len, alt_len); */
/* } */
/* qual */
token = strsep(&cur, delim);
if(token == NULL) {
my_err("expected at least %d tokens per line\n", n_fix_header);
}
util_strncpy(vcf_info->qual, token, sizeof(vcf_info->qual));
/* filter */
token = strsep(&cur, delim);
if(token == NULL) {
my_err("expected at least %d tokens per line\n", n_fix_header);
}
util_strncpy(vcf_info->filter, token, sizeof(vcf_info->filter));
/* info */
token = strsep(&cur, delim);
if(token == NULL) {
my_err("expected at least %d tokens per line\n", n_fix_header);
}
util_strncpy(vcf_info->info, token, sizeof(vcf_info->info));
/* format */
token = strsep(&cur, delim);
if(token == NULL) {
my_err("expected at least %d tokens per line\n", n_fix_header);
}
util_strncpy(vcf_info->format, token, sizeof(vcf_info->format));
/* now parse haplotypes and/or genotype likelihoods */
if(geno_probs && haplotypes) {
char *cur_copy;
/* Both genotype probs and haplotypes requested.
* Need to copy string because it is modified
* by the tokenizing in the parsing functions.
*
* This could be made more efficient by doing the parsing
* of both types of data at same time
*/
cur_copy = my_malloc(strlen(cur)+1);
strcpy(cur_copy, cur);
vcf_parse_geno_probs(vcf_info, geno_probs, cur_copy);
my_free(cur_copy);
vcf_parse_haplotypes(vcf_info, haplotypes, cur);
} else if(geno_probs) {
vcf_parse_geno_probs(vcf_info, geno_probs, cur);
} else if(haplotypes) {
vcf_parse_haplotypes(vcf_info, haplotypes, cur);
}
/* my_free(line); */
return 0;
}
