static int
utest_XPairId(ESL_ALPHABET *abc, char **as, ESL_DSQ **ax, int N)
{
double pid, pid2;
int nid, nid2;
int nres, nres2;
int dL, L;
int i,j;
/* Self comparison gives identity = 1. */
dL = esl_abc_dsqlen(ax[0]);
L = strlen(as[0]);
if (dL != L) abort();
if (esl_dst_XPairId(abc, ax[0], ax[0], &pid, &nid, &nres) != eslOK) abort();
if (pid != 1.0 || nid != L || nres > dL) abort();
/* So does 0,1 comparison */
if (esl_dst_XPairId(abc, ax[0], ax[1], &pid, &nid, &nres) != eslOK) abort();
if (pid != 1.0 || nid != L || nres > L) abort();
/* 0,2 comparison gives 0.0, 0 */
if (esl_dst_XPairId(abc, ax[0], ax[2], &pid, &nid, &nres) != eslOK) abort();
if (pid != 0.0 || nid != 0 || nres > L) abort();
/* remaining comparisons shouldn't fail, and should be identical to text mode */
for (i = 3; i < N; i++)
for (j = i; j < N; j++)
{
if (esl_dst_XPairId(abc, ax[i], ax[j], &pid, &nid, &nres) != eslOK) abort();
if (esl_dst_CPairId(as[i], as[j], &pid2, &nid2, &nres2) != eslOK) abort();
if (pid < 0. || pid > 1. || nid < 0 || nid > L || nres > L) abort();
if (pid != pid2 || nid != nid2 || nres != nres2) abort();
}
/* API should accept NULL for return values */
if (esl_dst_XPairId(abc, ax[0], ax[0], NULL, NULL, NULL) != eslOK) abort();
return eslOK;
}
/* Function: esl_msashuffle_XQRNA()
* Synopsis: Gap-preserving column shuffle of a digital pairwise alignment.
* Incept: SRE, Tue Jan 22 09:09:52 2008 [Market Street Cafe, Leesburg]
*
* Purpose: Shuffle a digital pairwise alignment <x>,<y> while
* preserving the position of gaps, where both sequences are
* in digital alphabet <abc>, using the random number
* generator <r>. Return the shuffled alignment in <xs>,
* <ys>. Caller provides allocated space for <xs> and <ys>
* for at least the same length of <x>,<y>.
*
* Works by doing three separate
* shuffles, of (1) columns with residues in both
* <x> and <y>, (2) columns with residue in <x> and gap in <y>,
* and (3) columns with gap in <x> and residue in <y>.
*
* <xs>,<x> and <ys>,<y> may be identical: that is, to shuffle
* an alignment "in place", destroying the original
* alignment, just call <esl_msashuffle_XQRNA(r, abc, x,y,x,y)>.
*
* Returns: <eslOK> on success, and the shuffled alignment is
* returned in <xs>, <ys>.
*
* Throws: <eslEMEM> on allocation failure.
*/
int
esl_msashuffle_XQRNA(ESL_RANDOMNESS *r, ESL_ALPHABET *abc, ESL_DSQ *x, ESL_DSQ *y, ESL_DSQ *xs, ESL_DSQ *ys)
{
int L;
int *xycol = NULL;
int *xcol = NULL;
int *ycol = NULL;
int nxy, nx, ny;
int i;
int pos, c;
char xsym, ysym;
int status;
L = esl_abc_dsqlen(x);
if (esl_abc_dsqlen(y) != L) ESL_XEXCEPTION(eslEINVAL, "sequences of different lengths in qrna shuffle");
if (xs != x) esl_abc_dsqcpy(x, L, xs);
if (ys != y) esl_abc_dsqcpy(y, L, ys);
/* First, construct three arrays containing lists of the column positions
* of the three types of columns. (If a column contains gaps in both x and y,
* we've already simply copied it to the shuffled sequence.)
*/
ESL_ALLOC(xycol, sizeof(int) * L);
ESL_ALLOC(xcol, sizeof(int) * L);
ESL_ALLOC(ycol, sizeof(int) * L);
nxy = nx = ny = 0;
for (i = 1; i <= L; i++)
{
if ( esl_abc_XIsGap(abc, x[i]) && esl_abc_XIsGap(abc, y[i])) { continue; }
else if (! esl_abc_XIsGap(abc, x[i]) && ! esl_abc_XIsGap(abc, y[i])) { xycol[nxy] = i; nxy++; }
else if ( esl_abc_XIsGap(abc, x[i])) { ycol[ny] = i; ny++; }
else if ( esl_abc_XIsGap(abc, y[i])) { xcol[nx] = i; nx++; }
}
/* Second, shuffle the sequences indirectly, via shuffling these arrays.
* Yow, careful with those indices, and with order of the statements...
*/
for (; nxy > 1; nxy--) {
pos = esl_rnd_Roll(r, nxy);
xsym = xs[xycol[pos]]; ysym = ys[xycol[pos]]; c = xycol[pos];
xs[xycol[pos]] = xs[xycol[nxy-1]]; ys[xycol[pos]] = ys[xycol[nxy-1]]; xycol[pos] = xycol[nxy-1];
xs[xycol[nxy-1]] = xsym; ys[xycol[nxy-1]] = ysym; xycol[pos] = xycol[nxy-1];
}
for (; nx > 1; nx--) {
pos = esl_rnd_Roll(r, nx);
xsym = xs[xcol[pos]]; ysym = ys[xcol[pos]]; c = xcol[pos];
xs[xcol[pos]] = xs[xcol[nx-1]]; ys[xcol[pos]] = ys[xcol[nx-1]]; xcol[pos] = xcol[nx-1];
xs[xcol[nx-1]] = xsym; ys[xcol[nx-1]] = ysym; xcol[nx-1] = c;
}
for (; ny > 1; ny--) {
pos = esl_rnd_Roll(r, ny);
xsym = xs[ycol[pos]]; ysym = ys[ycol[pos]]; c = ycol[pos];
xs[ycol[pos]] = xs[ycol[ny-1]]; ys[ycol[pos]] = ys[ycol[ny-1]]; ycol[pos] = ycol[ny-1];
xs[ycol[ny-1]] = xsym; ys[ycol[ny-1]] = ysym; ycol[ny-1] = c;
}
free(xycol); free(xcol); free(ycol);
return eslOK;
ERROR:
if (xycol != NULL) free(xycol);
if (xcol != NULL) free(xcol);
if (ycol != NULL) free(ycol);
return status;
}
