Traversing a Fatsa file in C/C++ - c++

Question

I'm looking to write a program in C/C++ to traverse a Fasta file formatted like:
>ID and header information
SEQUENCE1
>ID and header information
SEQUENCE2
and so on
in order to find all unique sequences (check if subset of any other sequence)
and write unique sequences (and all headers) to an output file.
My approach was:
Copy all sequences to an array/list at the beginning (more efficient way to do this?)
Grab header, append it to output file, compare sequence for that header to everything in the list/array. If unique, write it under the header, if not delete it.
However, I'm a little unsure as to how to approach reading the lines in properly. I need to read the top line for the header, and then "return?" to the next line to read the sequence. Sometimes the sequence spans more then two lines, so would I use > (from the example above) as a delimiter? If I use C++, I imagine I'd use iostreams to accomplish this?
If anybody could give me a nudge in the right direction as to how I would want to read the information I need to manipulate/how to carry out the comparison, it'd be greatly appreciated.

Answer 1

First, rather than write your own FASTA reading routine you probably want to use something that alrady exists, for example, see: http://lh3lh3.users.sourceforge.net/parsefastq.shtml
Internally you'll have the sequences without newlines and that is probably helpful. I think the simplest approach from a high level is
loop over fasta and write out sequences to a file
sort that file
with the sorted file it becomes easier to pick out subsequences so write a program to find the "unique ids"
Using the unique id's go back to the original fasta and get whatever additional information you need.

Answer 2

Your approach is usable. Below is an implementation of it.
However, I'm a little unsure as to how to approach reading the lines
in properly. ... Sometimes the sequence spans more then two lines, so would I use > (from the example above) as a delimiter?
That's right; in addition, there's just the EOF which has to be checked.
I wrote the function getd() for that, which reads a single-line description or concatenated lines of sequence data and returns a pointer to the string it allocated.
#include <stdio.h>
#include <stdlib.h>
#include <string.h>
char *getd()
{
char *s = NULL, *eol;
size_t i = 0, size = 1; // 1 for null character
int c;
#define MAXLINE 80 // recommended max. line length; longer lines are okay
do // read single-line description or concatenated lines of sequence data
{
do // read a line (until '\n')
{
s = realloc(s, size += MAXLINE+1); // +1 for newline character
if (!s) puts("out of memory"), exit(1);
if (!fgets(s+i, size, stdin)) break;
eol = strchr(s+i, '\n');
i += MAXLINE+1;
} while (!eol);
if (!i) { free(s); return NULL; } // nothing read
if (*s == '>') return s; // single-line description
i = eol-s;
ungetc(c = getchar(), stdin); // peek at next character
} while (c != '>' && c != EOF);
return s;
}
int main()
{
char *s;
struct seq { char *head, *data; } *seq = NULL;
int n = 0, i, j;
while (s = getd())
if (*s == '>')
{ // new sequence: make room, store header
seq = realloc(seq, ++n * sizeof *seq);
if (!seq) puts("out of memory"), exit(1);
seq[n-1] = (struct seq){ s, "" };
}
else
if (n) // store sequence data if at least one header present
seq[n-1].data = s;
for (i = 0; i < n; ++i)
{
const int max = 70; // reformat output data to that line length max.
printf("%s", seq[i].head);
for (s = seq[i].data, j = 0; j < n; ++j)
if (j != i) // compare sequence to the others, delete if not unique
if (strstr(seq[j].data, s)) { s = seq[i].data = ""; break; }
for (; strlen(s) > max && s[max] != '\n'; s += max)
printf("%.*s\n", max, s);
printf("%s", s);
}
}