boolean testInsertRestoreBIG (tree *tr, nodeptr p, nodeptr q)
{
if(Thorough)
{
if (! insertBIG(tr, p, q, tr->numBranches)) return FALSE;
evaluateGeneric(tr, p->next->next);
}
else
{
if (! insertRestoreBIG(tr, p, q)) return FALSE;
{
nodeptr x, y;
x = p->next->next;
y = p->back;
if(! isTip(x->number, tr->rdta->numsp) && isTip(y->number, tr->rdta->numsp))
{
while ((! x->x))
{
if (! (x->x))
newviewGeneric(tr, x);
}
}
if(isTip(x->number, tr->rdta->numsp) && !isTip(y->number, tr->rdta->numsp))
{
while ((! y->x))
{
if (! (y->x))
newviewGeneric(tr, y);
}
}
if(!isTip(x->number, tr->rdta->numsp) && !isTip(y->number, tr->rdta->numsp))
{
while ((! x->x) || (! y->x))
{
if (! (x->x))
newviewGeneric(tr, x);
if (! (y->x))
newviewGeneric(tr, y);
}
}
}
tr->likelihood = tr->endLH;
}
return TRUE;
}
reset_stNNI(state *s)
{
nodeptr p, q;
p = s->p;
q = p->back;
nodeptr pb1, pb2;
pb1 = p->next->back;
pb2 = p->next->next->back;
nodeptr qb1, qb2;
qb1 = q->next->back;
qb2 = q->next->next->back;
/* whichNNI 1*/
if(s->whichNNI == 1)
{
hookup(p, q, q->z_tmp, s->tr->numBranches);
hookup(p->next, qb1, p->next->z_tmp, s->tr->numBranches);
hookup(p->next->next, pb2, p->next->next->z_tmp, s->tr->numBranches);
hookup(q->next, pb1, q->next->z_tmp, s->tr->numBranches);
hookup(q->next->next, qb2, q->next->next->z_tmp, s->tr->numBranches);
}
else if(s->whichNNI == 2)
{
hookup(p, q, p->z_tmp, s->tr->numBranches);
hookup(p->next, pb1, p->next->z_tmp, s->tr->numBranches);
hookup(p->next->next, qb1, p->next->next->z_tmp, s->tr->numBranches);
hookup(q->next, pb2, q->next->z_tmp, s->tr->numBranches);
hookup(q->next->next, qb2, q->next->next->z_tmp, s->tr->numBranches);
}
else
{
assert(s->whichNNI == 0);
hookup(p, q, p->z_tmp, s->tr->numBranches);
hookup(p->next, pb1, p->next->z_tmp, s->tr->numBranches);
hookup(p->next->next, pb2, p->next->next->z_tmp, s->tr->numBranches);
hookup(q->next, qb1, q->next->z_tmp, s->tr->numBranches);
hookup(q->next->next, qb2, q->next->next->z_tmp, s->tr->numBranches);
}
/*
reset_branch_length(p, s->tr->numBranches);
reset_branch_length(p->next, s->tr->numBranches);
reset_branch_length(p->next->next, s->tr->numBranches);
reset_branch_length(q->next, s->tr->numBranches);
reset_branch_length(q->next->next, s->tr->numBranches);
*/
newviewGeneric(s->tr, p, FALSE);
newviewGeneric(s->tr, q, FALSE);
}
void treeEvaluateProgressive(tree *tr)
{
int
i, k;
tr->branchCounter = 0;
tr->numberOfBranches = 2 * tr->mxtips - 3;
tr->bInf = (branchInfo*)rax_malloc(tr->numberOfBranches * sizeof(branchInfo));
setupBranches(tr, tr->start->back, tr->bInf);
assert(tr->branchCounter == tr->numberOfBranches);
for(i = 0; i < tr->numBranches; i++)
tr->partitionConverged[i] = FALSE;
for(i = 0; i < 10; i++)
{
for(k = 0; k < tr->numberOfBranches; k++)
{
update(tr, tr->bInf[k].oP);
newviewGeneric(tr, tr->bInf[k].oP);
}
evaluateGenericInitrav(tr, tr->start);
printf("It %d %f \n", i, tr->likelihood);
}
}
nodeptr removeNodeRestoreBIG (tree *tr, nodeptr p)
{
nodeptr q, r;
q = p->next->back;
r = p->next->next->back;
newviewGeneric(tr, q);
newviewGeneric(tr, r);
hookup(q, r, tr->currentZQR, tr->numBranches);
p->next->next->back = p->next->back = (node *) NULL;
return q;
}
static void resetSimpleNodeProposal(state * instate)
{
/* prune the insertion */
hookup(instate->q, instate->r, instate->qz, instate->tr->numBranches);
instate->p->next->next->back = instate->p->next->back = (nodeptr) NULL;
/* insert the pruned tree in its original node */
hookup(instate->p->next, instate->nb, instate->nbz, instate->tr->numBranches);
hookup(instate->p->next->next, instate->nnb, instate->nnbz, instate->tr->numBranches);
newviewGeneric(instate->tr, instate->p, FALSE);
}
static double testInsertThorough(tree *tr, nodeptr r, nodeptr q, boolean useVector)
{
double
result,
qz[NUM_BRANCHES],
z[NUM_BRANCHES];
nodeptr
x = q->back,
s = r->back;
int
j;
for(j = 0; j < tr->numBranches; j++)
{
qz[j] = q->z[j];
z[j] = sqrt(qz[j]);
if(z[j] < zmin)
z[j] = zmin;
if(z[j] > zmax)
z[j] = zmax;
}
hookup(r->next, q, z, tr->numBranches);
hookup(r->next->next, x, z, tr->numBranches);
hookupDefault(r, s, tr->numBranches);
newviewGeneric(tr, r);
localSmooth(tr, r, smoothings);
if(useVector)
result = evaluateGenericVector(tr, r);
else
result = evaluateGeneric(tr, r);
hookup(q, x, qz, tr->numBranches);
r->next->next->back = r->next->back = (nodeptr) NULL;
return result;
}
boolean initrav (tree *tr, nodeptr p)
{
nodeptr q;
if (!isTip(p->number, tr->rdta->numsp))
{
q = p->next;
do
{
if (! initrav(tr, q->back)) return FALSE;
q = q->next;
}
while (q != p);
newviewGeneric(tr, p);
}
return TRUE;
}
boolean smooth (tree *tr, nodeptr p)
{
nodeptr q;
if (! update(tr, p)) return FALSE; /* Adjust branch */
if (! isTip(p->number, tr->rdta->numsp))
{ /* Adjust descendants */
q = p->next;
while (q != p)
{
if (! smooth(tr, q->back)) return FALSE;
q = q->next;
}
if(tr->multiBranch)
newviewGenericMasked(tr, p);
else
newviewGeneric(tr, p);
}
return TRUE;
}
static void traverse_branches(nodeptr p, int *count, state * s, boolean resetBL)
{
nodeptr q;
//printf("current BL at %db%d: %f\n", p->number, p->back->number, p->z[0]);
if(resetBL)
reset_branch_length(p, s->tr->numBranches);
else//can allow for other methods later
set_branch_length_sliding_window(p, s->tr->numBranches, s, TRUE);
*count += 1;
if (! isTip(p->number, s->tr->mxtips))
{ /* Adjust descendants */
q = p->next;
while (q != p)
{
traverse_branches(q->back, count, s, resetBL);
q = q->next;
}
newviewGeneric(s->tr, p, FALSE); // not sure if we need this
}
}
boolean smoothRegion (tree *tr, nodeptr p, int region)
{
nodeptr q;
if (! update(tr, p)) return FALSE; /* Adjust branch */
if(region > 0)
{
if (!isTip(p->number, tr->rdta->numsp))
{
q = p->next;
while (q != p)
{
if (! smoothRegion(tr, q->back, --region)) return FALSE;
q = q->next;
}
newviewGeneric(tr, p);
}
}
return TRUE;
}
boolean insertRestoreBIG (tree *tr, nodeptr p, nodeptr q)
{
nodeptr r, s;
r = q->back;
s = p->back;
if(Thorough)
{
hookup(p->next, q, tr->currentLZQ, tr->numBranches);
hookup(p->next->next, r, tr->currentLZR, tr->numBranches);
hookup(p, s, tr->currentLZS, tr->numBranches);
}
else
{
double z[NUM_BRANCHES];
int i;
for(i = 0; i < tr->numBranches; i++)
{
double zz;
zz = sqrt(q->z[i]);
if(zz < zmin)
zz = zmin;
if(zz > zmax)
zz = zmax;
z[i] = zz;
}
hookup(p->next, q, z, tr->numBranches);
hookup(p->next->next, r, z, tr->numBranches);
}
newviewGeneric(tr, p);
return TRUE;
}
static boolean qsmoothLocal(tree *tr, nodeptr p, int n)
{
nodeptr q;
if(n == 0)
return TRUE;
else
{
if (! qupdate(tr, p)) return FALSE; /* Adjust branch */
if (!isTip(p->number, tr->rdta->numsp))
{ /* Adjust descendants */
q = p->next;
while (q != p)
{
if (! qsmoothLocal(tr, q->back, n - 1)) return FALSE;
q = q->next;
}
newviewGeneric(tr, p);
}
return TRUE;
}
}
static void insertFast (tree *tr, nodeptr p, nodeptr q, int numBranches)
{
nodeptr r, s;
int i;
r = q->back;
s = p->back;
for(i = 0; i < numBranches; i++)
tr->lzi[i] = q->z[i];
if(Thorough)
{
double zqr[NUM_BRANCHES], zqs[NUM_BRANCHES], zrs[NUM_BRANCHES], lzqr, lzqs, lzrs, lzsum, lzq, lzr, lzs, lzmax;
double defaultArray[NUM_BRANCHES];
double e1[NUM_BRANCHES], e2[NUM_BRANCHES], e3[NUM_BRANCHES];
double *qz;
qz = q->z;
for(i = 0; i < numBranches; i++)
defaultArray[i] = defaultz;
makenewzGeneric(tr, q, r, qz, iterations, zqr, FALSE);
makenewzGeneric(tr, q, s, defaultArray, iterations, zqs, FALSE);
makenewzGeneric(tr, r, s, defaultArray, iterations, zrs, FALSE);
for(i = 0; i < numBranches; i++)
{
lzqr = (zqr[i] > zmin) ? log(zqr[i]) : log(zmin);
lzqs = (zqs[i] > zmin) ? log(zqs[i]) : log(zmin);
lzrs = (zrs[i] > zmin) ? log(zrs[i]) : log(zmin);
lzsum = 0.5 * (lzqr + lzqs + lzrs);
lzq = lzsum - lzrs;
lzr = lzsum - lzqs;
lzs = lzsum - lzqr;
lzmax = log(zmax);
if (lzq > lzmax) {lzq = lzmax; lzr = lzqr; lzs = lzqs;}
else if (lzr > lzmax) {lzr = lzmax; lzq = lzqr; lzs = lzrs;}
else if (lzs > lzmax) {lzs = lzmax; lzq = lzqs; lzr = lzrs;}
e1[i] = exp(lzq);
e2[i] = exp(lzr);
e3[i] = exp(lzs);
}
hookup(p->next, q, e1, numBranches);
hookup(p->next->next, r, e2, numBranches);
hookup(p, s, e3, numBranches);
}
else
{
double z[NUM_BRANCHES];
for(i = 0; i < numBranches; i++)
{
z[i] = sqrt(q->z[i]);
if(z[i] < zmin)
z[i] = zmin;
if(z[i] > zmax)
z[i] = zmax;
}
hookup(p->next, q, z, tr->numBranches);
hookup(p->next->next, r, z, tr->numBranches);
}
newviewGeneric(tr, p);
if(Thorough)
{
localSmooth(tr, p, smoothings);
for(i = 0; i < numBranches; i++)
{
tr->lzq[i] = p->next->z[i];
tr->lzr[i] = p->next->next->z[i];
tr->lzs[i] = p->z[i];
}
}
}
int main(int argc, char * argv[])
{
tree * tr;
if (argc != 2)
{
fprintf (stderr, "syntax: %s [binary-alignment-file]\n", argv[0]);
return (1);
}
tr = (tree *)malloc(sizeof(tree));
/* read the binary input, setup tree, initialize model with alignment */
read_msa(tr,argv[1]);
tr->randomNumberSeed = 665;
makeRandomTree(tr);
printf("Number of taxa: %d\n", tr->mxtips);
printf("Number of partitions: %d\n", tr->NumberOfModels);
/* compute the LH of the full tree */
printf ("Virtual root: %d\n", tr->start->number);
evaluateGeneric(tr, tr->start, TRUE);
printf("Likelihood: %f\n", tr->likelihood);
/* 8 rounds of branch length optimization */
smoothTree(tr, 1);
evaluateGeneric(tr, tr->start, TRUE);
printf("Likelihood after branch length optimization: %.20f\n", tr->likelihood);
/* Now we show how to find a particular LH vector for a node */
int i;
int node_number = tr->mxtips + 1;
nodeptr p = tr->nodep[node_number];
printf("Pointing to node %d\n", p->number);
/* Fix as VR */
newviewGeneric(tr, p, FALSE);
newviewGeneric(tr, p->back, FALSE);
evaluateGeneric(tr, p, FALSE);
printf("Likelihood : %.f\n", tr->likelihood);
printf("Make a copy of LH vector for node %d\n", p->number);
likelihood_vector *vector = copy_likelihood_vectors(tr, p);
for(i=0; i<vector->num_partitions; i++)
printf("Partition %d requires %d bytes\n", i, (int)vector->partition_sizes[i]);
/* Check we have the same vector in both tree and copied one */
assert(same_vector(tr, p, vector));
/* Now force the p to get a new value (generally branch lengths are NOT updated like this) */
/* This is just an example to show usage (for fast NNI eval), manually updating vectors is not recommended! */
printf("bl : %.40f\n", p->next->z[0]);
p->next->z[0] = p->next->back->z[0] = zmin;
printf("bl : %.40f\n", p->next->z[0]);
newviewGeneric(tr, p, FALSE);
assert(!same_vector(tr, p, vector));
evaluateGeneric(tr, p, FALSE);
printf("Likelihood : %f\n", tr->likelihood);
restore_vector(tr, p, vector);
assert(same_vector(tr, p, vector));
evaluateGeneric(tr, p, FALSE);
printf("Likelihood after manually restoring the vector : %f\n", tr->likelihood);
free_likelihood_vector(vector);
/* Pick an inner branch */
printf("numBranches %d \n", tr->numBranches);
//tr->numBranches = 1;
p = tr->nodep[tr->mxtips + 1];
int partition_id = 0; /* single partition */
double bl = get_branch_length(tr, p, partition_id);
printf("z value: %f , bl value %f\n", p->z[partition_id], bl);
/* set the bl to 2.5 */
double new_bl = 2.5;
set_branch_length(tr, p, partition_id, new_bl);
printf("Changed BL to %f\n", new_bl);
printf("new z value: %f , new bl value %f\n", p->z[partition_id], get_branch_length(tr, p, partition_id));
/* set back to original */
printf("Changed to previous BL\n");
set_branch_length(tr, p, partition_id, bl);
printf("new z value: %f , new bl value %f\n", p->z[partition_id], get_branch_length(tr, p, partition_id));
return (0);
}
static double linearSPRs(tree *tr, int radius, boolean veryFast)
{
int
numberOfSubtrees = (tr->mxtips - 2) * 3,
count = 0,
k,
i;
double
fourScores[4];
nodeptr
*ptr = (nodeptr *)rax_malloc(sizeof(nodeptr) * numberOfSubtrees);
fourLikelihoods
*fourLi = (fourLikelihoods *)rax_malloc(sizeof(fourLikelihoods) * 4);
insertions
*ins = (insertions*)rax_malloc(sizeof(insertions));
ins->count = 0;
ins->maxCount = 2048;
ins->s = (scores *)rax_malloc(sizeof(scores) * ins->maxCount);
/* recursively compute the roots of all subtrees in the current tree
and store them in ptr */
getSubtreeRoots(tr->start->back, ptr, &count, tr->mxtips);
assert(count == numberOfSubtrees);
tr->startLH = tr->endLH = tr->likelihood;
/* loop over subtrees, i.e., execute a full SPR cycle */
for(i = 0; i < numberOfSubtrees; i++)
{
nodeptr
p = ptr[i],
p1 = p->next->back,
p2 = p->next->next->back;
double
p1z[NUM_BRANCHES],
p2z[NUM_BRANCHES];
ins->count = 0;
/*printf("Node %d %d\n", p->number, i);*/
tr->bestOfNode = unlikely;
for(k = 0; k < 4; k++)
fourScores[k] = unlikely;
assert(!isTip(p->number, tr->rdta->numsp));
if(!isTip(p1->number, tr->rdta->numsp) || !isTip(p2->number, tr->rdta->numsp))
{
double
max = unlikely;
int
maxInt = -1;
for(k = 0; k < tr->numBranches; k++)
{
p1z[k] = p1->z[k];
p2z[k] = p2->z[k];
}
/* remove the current subtree */
removeNodeBIG(tr, p, tr->numBranches);
/* pre score with fast insertions */
if(veryFast)
Thorough = 1;
else
Thorough = 0;
if (!isTip(p1->number, tr->rdta->numsp))
{
fourScores[0] = testInsertFast(tr, p, p1->next->back, ins, veryFast);
fourScores[1] = testInsertFast(tr, p, p1->next->next->back, ins, veryFast);
}
if (!isTip(p2->number, tr->rdta->numsp))
{
fourScores[2] = testInsertFast(tr, p, p2->next->back, ins, veryFast);
fourScores[3] = testInsertFast(tr, p, p2->next->next->back, ins, veryFast);
}
if(veryFast)
Thorough = 1;
else
Thorough = 0;
/* find the most promising direction */
if(!veryFast)
{
int
j = 0,
validEntries = 0;
double
lmax = unlikely,
posterior = 0.0;
for(k = 0; k < 4; k++)
{
fourLi[k].direction = k;
fourLi[k].likelihood = fourScores[k];
}
qsort(fourLi, 4, sizeof(fourLikelihoods), fourCompare);
for(k = 0; k < 4; k++)
if(fourLi[k].likelihood > unlikely)
validEntries++;
lmax = fourLi[0].likelihood;
while(posterior <= POSTERIOR_THRESHOLD && j < validEntries)
{
double
all = 0.0,
prob = 0.0;
for(k = 0; k < validEntries; k++)
all += exp(fourLi[k].likelihood - lmax);
posterior += (prob = (exp(fourLi[j].likelihood - lmax) / all));
switch(fourLi[j].direction)
{
case 0:
insertBeyond(tr, p, p1->next->back, radius, ins, veryFast);
break;
case 1:
insertBeyond(tr, p, p1->next->next->back, radius, ins, veryFast);
break;
case 2:
insertBeyond(tr, p, p2->next->back, radius, ins, veryFast);
break;
case 3:
insertBeyond(tr, p, p2->next->next->back, radius, ins, veryFast);
break;
default:
assert(0);
}
j++;
}
qsort(ins->s, ins->count, sizeof(scores), scoreCompare);
Thorough = 1;
for(k = 0; k < MIN(ins->count, 20); k++)
testInsertCandidates(tr, p, ins->s[k].p);
}
else
{
Thorough = 1;
for(k = 0; k < 4; k++)
{
if(max < fourScores[k])
{
max = fourScores[k];
maxInt = k;
}
}
/* descend into this direction and re-insert subtree there */
if(maxInt >= 0)
{
switch(maxInt)
{
case 0:
insertBeyond(tr, p, p1->next->back, radius, ins, veryFast);
break;
case 1:
insertBeyond(tr, p, p1->next->next->back, radius, ins, veryFast);
break;
case 2:
insertBeyond(tr, p, p2->next->back, radius, ins, veryFast);
break;
case 3:
insertBeyond(tr, p, p2->next->next->back, radius, ins, veryFast);
break;
default:
assert(0);
}
}
}
/* repair branch and reconnect subtree to its original position from which it was pruned */
hookup(p->next, p1, p1z, tr->numBranches);
hookup(p->next->next, p2, p2z, tr->numBranches);
/* repair likelihood vectors */
newviewGeneric(tr, p);
/* if the rearrangement of subtree rooted at p yielded a better likelihood score
restore the altered topology and use it from now on
*/
if(tr->endLH > tr->startLH)
{
restoreTreeFast(tr);
tr->startLH = tr->endLH = tr->likelihood;
}
}
}
return tr->startLH;
}
static void doNNIs(tree *tr, nodeptr p, double *lhVectors[3], boolean shSupport, int *interchanges, int *innerBranches,
double *pqz_0, double *pz1_0, double *pz2_0, double *qz1_0, double *qz2_0, double *pqz_1, double *pz1_1, double *pz2_1,
double *qz1_1, double *qz2_1, double *pqz_2, double *pz1_2, double *pz2_2, double *qz1_2, double *qz2_2)
{
nodeptr
q = p->back,
pb1 = p->next->back,
pb2 = p->next->next->back;
assert(!isTip(p->number, tr->mxtips));
if(!isTip(q->number, tr->mxtips))
{
int
whichNNI = 0;
nodeptr
qb1 = q->next->back,
qb2 = q->next->next->back;
double
lh[3];
*innerBranches = *innerBranches + 1;
nniSmooth(tr, p, 16);
if(shSupport)
{
evaluateGenericVector(tr, p);
memcpy(lhVectors[0], tr->perSiteLL, sizeof(double) * tr->cdta->endsite);
}
else
evaluateGeneric(tr, p);
lh[0] = tr->likelihood;
storeBranches(tr, p, pqz_0, pz1_0, pz2_0, qz1_0, qz2_0);
/*******************************************/
hookup(p, q, pqz_0, tr->numBranches);
hookup(p->next, qb1, qz1_0, tr->numBranches);
hookup(p->next->next, pb2, pz2_0, tr->numBranches);
hookup(q->next, pb1, pz1_0, tr->numBranches);
hookup(q->next->next, qb2, qz2_0, tr->numBranches);
newviewGeneric(tr, p);
newviewGeneric(tr, p->back);
nniSmooth(tr, p, 16);
if(shSupport)
{
evaluateGenericVector(tr, p);
memcpy(lhVectors[1], tr->perSiteLL, sizeof(double) * tr->cdta->endsite);
}
else
evaluateGeneric(tr, p);
lh[1] = tr->likelihood;
storeBranches(tr, p, pqz_1, pz1_1, pz2_1, qz1_1, qz2_1);
if(lh[1] > lh[0])
whichNNI = 1;
/*******************************************/
hookup(p, q, pqz_0, tr->numBranches);
hookup(p->next, qb1, qz1_0, tr->numBranches);
hookup(p->next->next, pb1, pz1_0, tr->numBranches);
hookup(q->next, pb2, pz2_0, tr->numBranches);
hookup(q->next->next, qb2, qz2_0, tr->numBranches);
newviewGeneric(tr, p);
newviewGeneric(tr, p->back);
nniSmooth(tr, p, 16);
if(shSupport)
{
evaluateGenericVector(tr, p);
memcpy(lhVectors[2], tr->perSiteLL, sizeof(double) * tr->cdta->endsite);
}
else
evaluateGeneric(tr, p);
lh[2] = tr->likelihood;
storeBranches(tr, p, pqz_2, pz1_2, pz2_2, qz1_2, qz2_2);
if(lh[2] > lh[0] && lh[2] > lh[1])
whichNNI = 2;
/*******************************************/
if(shSupport)
whichNNI = 0;
switch(whichNNI)
{
case 0:
hookup(p, q, pqz_0, tr->numBranches);
hookup(p->next, pb1, pz1_0, tr->numBranches);
hookup(p->next->next, pb2, pz2_0, tr->numBranches);
hookup(q->next, qb1, qz1_0, tr->numBranches);
hookup(q->next->next, qb2, qz2_0, tr->numBranches);
break;
case 1:
hookup(p, q, pqz_1, tr->numBranches);
hookup(p->next, qb1, pz1_1, tr->numBranches);
hookup(p->next->next, pb2, pz2_1, tr->numBranches);
hookup(q->next, pb1, qz1_1, tr->numBranches);
hookup(q->next->next, qb2, qz2_1, tr->numBranches);
break;
case 2:
hookup(p, q, pqz_2, tr->numBranches);
hookup(p->next, qb1, pz1_2, tr->numBranches);
hookup(p->next->next, pb1, pz2_2, tr->numBranches);
hookup(q->next, pb2, qz1_2, tr->numBranches);
hookup(q->next->next, qb2, qz2_2, tr->numBranches);
break;
default:
assert(0);
}
newviewGeneric(tr, p);
newviewGeneric(tr, q);
if(whichNNI > 0)
*interchanges = *interchanges + 1;
if(shSupport)
p->bInf->support = SHSupport(tr->cdta->endsite, 1000, tr->resample, lh, lhVectors);
}
if(!isTip(pb1->number, tr->mxtips))
doNNIs(tr, pb1, lhVectors, shSupport, interchanges, innerBranches,
pqz_0, pz1_0, pz2_0, qz1_0, qz2_0, pqz_1, pz1_1, pz2_1,
qz1_1, qz2_1, pqz_2, pz1_2, pz2_2, qz1_2, qz2_2);
if(!isTip(pb2->number, tr->mxtips))
doNNIs(tr, pb2, lhVectors, shSupport, interchanges, innerBranches,
pqz_0, pz1_0, pz2_0, qz1_0, qz2_0, pqz_1, pz1_1, pz2_1,
qz1_1, qz2_1, pqz_2, pz1_2, pz2_2, qz1_2, qz2_2);
return;
}
static boolean stNNIproposal(state *s)
{
//s->newprior = 1;
s->bl_prior = 0;
int attempts = 0;
do{
s->p = selectRandomInnerSubtree(s->tr); /* TODOFER do this ad hoc for NNI requirements*/
if (++attempts > 500)
return FALSE;
}while(isTip(s->p->number, s->tr->mxtips) || isTip(s->p->back->number, s->tr->mxtips));
assert(!isTip(s->p->number, s->tr->mxtips));
nodeptr
p = s->p,
q = s->p->back,
pb1 = s->p->next->back,
pb2 = s->p->next->next->back;
assert(!isTip(q->number, s->tr->mxtips));
nodeptr
qb1 = q->next->back,
qb2 = q->next->next->back;
recordNNIBranchInfo(p, s->tr->numBranches);
/* do only one type of NNI, nni1 */
double randprop = (double)rand()/(double)RAND_MAX;
boolean changeBL = TRUE;
if (randprop < 1.0 / 3.0)
{
s->whichNNI = 1;
if(!changeBL)
{
hookup(p, q, p->z, s->tr->numBranches);
hookup(p->next, qb1, q->next->z, s->tr->numBranches);
hookup(p->next->next, pb2, p->next->next->z, s->tr->numBranches);
hookup(q->next, pb1, p->next->z, s->tr->numBranches);
hookup(q->next->next, qb2, q->next->next->z, s->tr->numBranches);
}
else
{
hookupBL(p, q, p, s);
hookupBL(p->next, qb1, q->next, s);
hookupBL(p->next->next, pb2, p->next->next, s);
hookupBL(q->next, pb1, p->next, s);
hookupBL(q->next->next, qb2, q->next->next, s);
}
}
else if (randprop < 2.0 / 3.0)
{
s->whichNNI = 2;
if(!changeBL)
{
hookup(p, q, p->z, s->tr->numBranches);
hookup(p->next, pb1, p->next->z, s->tr->numBranches);
hookup(p->next->next, qb1, q->next->z, s->tr->numBranches);
hookup(q->next, pb2, p->next->next->z, s->tr->numBranches);
hookup(q->next->next, qb2, q->next->next->z, s->tr->numBranches);
}
else
{
hookupBL(p, q, p, s);
hookupBL(p->next, pb1, p->next, s);
hookupBL(p->next->next, qb1, q->next, s);
hookupBL(q->next, pb2, p->next->next, s);
hookupBL(q->next->next, qb2, q->next->next, s);
}
}
else
{
/* change only the branch lengths */
s->whichNNI = 0;
if(changeBL)
{
/* do it like this for symmetry */
hookupBL(p, q, p, s);
hookupBL(p->next, pb1, p->next, s);
hookupBL(p->next->next, pb2, p->next->next, s);
hookupBL(q->next, qb1, q->next, s);
hookupBL(q->next->next, qb2, q->next->next, s);
}
}
newviewGeneric(s->tr, p, FALSE);
newviewGeneric(s->tr, p->back, FALSE);
evaluateGeneric(s->tr, p, FALSE);
return TRUE;
}
void computePlacementBias(tree *tr, analdef *adef)
{
int
windowSize = adef->slidingWindowSize,
k,
i,
tips,
numTraversalBranches = (2 * (tr->mxtips - 1)) - 3; /* compute number of branches into which we need to insert once we have removed a taxon */
char
fileName[1024];
FILE
*outFile;
/* data for each sliding window starting position */
positionData
*pd = (positionData *)malloc(sizeof(positionData) * (tr->cdta->endsite - windowSize));
double
*nodeDistances = (double*)calloc(tr->cdta->endsite, sizeof(double)), /* array to store node distnces ND for every sliding window position */
*distances = (double*)calloc(tr->cdta->endsite, sizeof(double)); /* array to store avg distances for every site */
strcpy(fileName, workdir);
strcat(fileName, "RAxML_SiteSpecificPlacementBias.");
strcat(fileName, run_id);
outFile = myfopen(fileName, "w");
printBothOpen("Likelihood of comprehensive tree %f\n\n", tr->likelihood);
if(windowSize > tr->cdta->endsite)
{
printBothOpen("The size of your sliding window is %d while the number of sites in the alignment is %d\n\n", windowSize, tr->cdta->endsite);
exit(-1);
}
if(windowSize >= (int)(0.9 * tr->cdta->endsite))
printBothOpen("WARNING: your sliding window of size %d is only slightly smaller than you alignment that has %d sites\n\n", windowSize, tr->cdta->endsite);
printBothOpen("Sliding window size: %d\n\n", windowSize);
/* prune and re-insert on tip at a time into all branches of the remaining tree */
for(tips = 1; tips <= tr->mxtips; tips++)
{
nodeptr
myStart,
p = tr->nodep[tips]->back, /* this is the node at which we are prunung */
p1 = p->next->back,
p2 = p->next->next->back;
double
pz[NUM_BRANCHES],
p1z[NUM_BRANCHES],
p2z[NUM_BRANCHES];
int
branchCounter = 0;
/* reset array values for this tip */
for(i = 0; i < tr->cdta->endsite; i++)
{
pd[i].lh = unlikely;
pd[i].p = (nodeptr)NULL;
}
/* store the three branch lengths adjacent to the position at which we prune */
for(i = 0; i < tr->numBranches; i++)
{
p1z[i] = p1->z[i];
p2z[i] = p2->z[i];
pz[i] = p->z[i];
}
/* prune the taxon, optimizing the branch between p1 and p2 */
removeNodeBIG(tr, p, tr->numBranches);
printBothOpen("Pruning taxon Number %d [%s]\n", tips, tr->nameList[tips]);
/* find any tip to start traversing the tree */
myStart = findAnyTip(p1, tr->mxtips);
/* insert taxon, compute likelihood and remove taxon again from all branches */
traverseBias(p, myStart->back, tr, &branchCounter, pd, windowSize);
assert(branchCounter == numTraversalBranches);
/* for every sliding window position calc ND to the true/correct position at p */
for(i = 0; i < tr->cdta->endsite - windowSize; i++)
nodeDistances[i] = getNodeDistance(p1, pd[i].p, tr->mxtips);
/* now analyze */
for(i = 0; i < tr->cdta->endsite; i++)
{
double
d = 0.0;
int
s = 0;
/*
check site position, i.e., doe we have windowSize data points available
or fewer because we are at the start or the end of the alignment
*/
/*
for each site just accumulate the node distances we have for all
sliding windows that passed over this site
*/
if(i < windowSize)
{
for(k = 0; k < i + 1; k++, s++)
d += nodeDistances[k];
}
else
{
if(i < tr->cdta->endsite - windowSize)
{
for(k = i - windowSize + 1; k <= i; k++, s++)
d += nodeDistances[k];
}
else
{
for(k = i - windowSize; k < (tr->cdta->endsite - windowSize); k++, s++)
d += nodeDistances[k + 1];
}
}
/*
now just divide the accumultaed ND distance by
the number of distances we have for this position and then add it to the acc
distances over all taxa.
I just realized that the version on which I did the tests
I sent to Simon I used
distances[i] = d / ((double)s);
instead of
distances[i] += d / ((double)s);
gamo tin poutana mou
*/
distances[i] += (d / ((double)s));
}
/*
re-connect taxon to its original position
*/
hookup(p->next, p1, p1z, tr->numBranches);
hookup(p->next->next, p2, p2z, tr->numBranches);
hookup(p, p->back, pz, tr->numBranches);
/*
fix likelihood vectors
*/
newviewGeneric(tr, p);
}
/*
now just compute the average ND over all taxa
*/
for(i = 0; i < tr->cdta->endsite; i++)
{
double
avg = distances[i] / ((double)tr->mxtips);
fprintf(outFile, "%d %f\n", i, avg);
}
printBothOpen("\nTime for EPA-based site-specific placement bias calculation: %f\n", gettime() - masterTime);
printBothOpen("Site-specific placement bias statistics written to file %s\n", fileName);
fclose(outFile);
exit(0);
}
int rearrangeBIG(tree *tr, nodeptr p, int mintrav, int maxtrav)
{
double p1z[NUM_BRANCHES], p2z[NUM_BRANCHES], q1z[NUM_BRANCHES], q2z[NUM_BRANCHES];
nodeptr p1, p2, q, q1, q2;
int mintrav2, i;
boolean doP = TRUE, doQ = TRUE;
if (maxtrav < 1 || mintrav > maxtrav) return 0;
q = p->back;
if(tr->constrained)
{
if(! tipHomogeneityChecker(tr, p->back, 0))
doP = FALSE;
if(! tipHomogeneityChecker(tr, q->back, 0))
doQ = FALSE;
if(doQ == FALSE && doP == FALSE)
return 0;
}
if (!isTip(p->number, tr->rdta->numsp) && doP)
{
p1 = p->next->back;
p2 = p->next->next->back;
if(!isTip(p1->number, tr->rdta->numsp) || !isTip(p2->number, tr->rdta->numsp))
{
for(i = 0; i < tr->numBranches; i++)
{
p1z[i] = p1->z[i];
p2z[i] = p2->z[i];
}
if (! removeNodeBIG(tr, p, tr->numBranches)) return badRear;
if (!isTip(p1->number, tr->rdta->numsp))
{
addTraverseBIG(tr, p, p1->next->back,
mintrav, maxtrav);
addTraverseBIG(tr, p, p1->next->next->back,
mintrav, maxtrav);
}
if (!isTip(p2->number, tr->rdta->numsp))
{
addTraverseBIG(tr, p, p2->next->back,
mintrav, maxtrav);
addTraverseBIG(tr, p, p2->next->next->back,
mintrav, maxtrav);
}
hookup(p->next, p1, p1z, tr->numBranches);
hookup(p->next->next, p2, p2z, tr->numBranches);
newviewGeneric(tr, p);
}
}
if (!isTip(q->number, tr->rdta->numsp) && maxtrav > 0 && doQ)
{
q1 = q->next->back;
q2 = q->next->next->back;
/*if (((!q1->tip) && (!q1->next->back->tip || !q1->next->next->back->tip)) ||
((!q2->tip) && (!q2->next->back->tip || !q2->next->next->back->tip))) */
if (
(
! isTip(q1->number, tr->rdta->numsp) &&
(! isTip(q1->next->back->number, tr->rdta->numsp) || ! isTip(q1->next->next->back->number, tr->rdta->numsp))
)
||
(
! isTip(q2->number, tr->rdta->numsp) &&
(! isTip(q2->next->back->number, tr->rdta->numsp) || ! isTip(q2->next->next->back->number, tr->rdta->numsp))
)
)
{
for(i = 0; i < tr->numBranches; i++)
{
q1z[i] = q1->z[i];
q2z[i] = q2->z[i];
}
if (! removeNodeBIG(tr, q, tr->numBranches)) return badRear;
mintrav2 = mintrav > 2 ? mintrav : 2;
if (/*! q1->tip*/ !isTip(q1->number, tr->rdta->numsp))
{
addTraverseBIG(tr, q, q1->next->back,
mintrav2 , maxtrav);
addTraverseBIG(tr, q, q1->next->next->back,
mintrav2 , maxtrav);
}
if (/*! q2->tip*/ ! isTip(q2->number, tr->rdta->numsp))
{
addTraverseBIG(tr, q, q2->next->back,
mintrav2 , maxtrav);
addTraverseBIG(tr, q, q2->next->next->back,
mintrav2 , maxtrav);
}
hookup(q->next, q1, q1z, tr->numBranches);
hookup(q->next->next, q2, q2z, tr->numBranches);
newviewGeneric(tr, q);
}
}
return 1;
}
static void smoothTreeRandom (tree *tr, int maxtimes)
{
int i, k, *perm;
tr->branchCounter = 0;
tr->numberOfBranches = 2 * tr->mxtips - 3;
tr->bInf = (branchInfo*)rax_malloc(tr->numberOfBranches * sizeof(branchInfo));
perm = (int*)rax_malloc(sizeof(int) * tr->numberOfBranches);
setupBranches(tr, tr->start->back, tr->bInf);
for(i = 0; i < tr->numBranches; i++)
tr->partitionConverged[i] = FALSE;
/*printf("%d \n", maxtimes);*/
while (--maxtimes >= 0)
{
for(i = 0; i < tr->numBranches; i++)
tr->partitionSmoothed[i] = TRUE;
/*printf("%d %d\n", maxtimes, tr->numberOfBranches);*/
for(k = 0; k < tr->numberOfBranches; k++)
perm[k] = k;
makePerm(perm, tr->numberOfBranches);
for(k = 0; k < tr->numberOfBranches; k++)
{
/*printf("%d Node %d\n", k, tr->bInf[k].oP->number);*/
update(tr, tr->bInf[perm[k]].oP);
newviewGeneric(tr, tr->bInf[perm[k]].oP);
}
/*if (! smooth(tr, p->back)) return FALSE;
if (!isTip(p->number, tr->rdta->numsp))
{
q = p->next;
while (q != p)
{
if (! smooth(tr, q->back)) return FALSE;
q = q->next;
}
}
*/
if (allSmoothed(tr))
break;
}
for(i = 0; i < tr->numBranches; i++)
tr->partitionConverged[i] = FALSE;
rax_free(tr->bInf);
rax_free(perm);
}
static void computeAncestralRec(tree *tr, nodeptr p, int *counter, FILE *probsFile, FILE *statesFile, boolean atRoot)
{
#ifdef _USE_PTHREADS
size_t
accumulatedOffset = 0;
#endif
int
model,
globalIndex = 0;
ancestralState
*a = (ancestralState *)malloc(sizeof(ancestralState) * tr->cdta->endsite),
*unsortedA = (ancestralState *)malloc(sizeof(ancestralState) * tr->rdta->sites);
if(!atRoot)
{
if(isTip(p->number, tr->mxtips))
return;
computeAncestralRec(tr, p->next->back, counter, probsFile, statesFile, atRoot);
computeAncestralRec(tr, p->next->next->back, counter, probsFile, statesFile, atRoot);
newviewGeneric(tr, p);
}
newviewGenericAncestral(tr, p, atRoot);
#ifdef _USE_PTHREADS
masterBarrier(THREAD_GATHER_ANCESTRAL, tr);
#endif
if(atRoot)
{
fprintf(probsFile, "ROOT\n");
fprintf(statesFile, "ROOT ");
}
else
{
fprintf(probsFile, "%d\n", p->number);
fprintf(statesFile, "%d ", p->number);
}
for(model = 0; model < tr->NumberOfModels; model++)
{
int
offset,
i,
width = tr->partitionData[model].upper - tr->partitionData[model].lower,
states = tr->partitionData[model].states;
#ifdef _USE_PTHREADS
double
*ancestral = tr->ancestralStates;
#else
double
*ancestral = tr->partitionData[model].sumBuffer;
#endif
if(tr->rateHetModel == CAT)
offset = 1;
else
offset = 4;
for(i = 0; i < width; i++, globalIndex++)
{
double
equal = 1.0 / (double)states,
max = -1.0;
boolean
approximatelyEqual = TRUE;
int
max_l = -1,
l;
char
c;
a[globalIndex].states = states;
a[globalIndex].probs = (double *)malloc(sizeof(double) * states);
for(l = 0; l < states; l++)
{
double
value = ancestral[offset * states * i + l];
if(value > max)
{
max = value;
max_l = l;
}
approximatelyEqual = approximatelyEqual && (ABS(equal - value) < 0.000001);
a[globalIndex].probs[l] = value;
}
if(approximatelyEqual)
c = '?';
else
c = getStateCharacter(tr->partitionData[model].dataType, max_l);
a[globalIndex].c = c;
}
#ifdef _USE_PTHREADS
accumulatedOffset += width * offset * states;
#endif
}
{
int
j,
k;
for(j = 0; j < tr->cdta->endsite; j++)
{
for(k = 0; k < tr->rdta->sites; k++)
if(j == tr->patternPosition[k])
{
int
sorted = j,
unsorted = tr->columnPosition[k] - 1;
unsortedA[unsorted].states = a[sorted].states;
unsortedA[unsorted].c = a[sorted].c;
unsortedA[unsorted].probs = (double*)malloc(sizeof(double) * unsortedA[unsorted].states);
memcpy(unsortedA[unsorted].probs, a[sorted].probs, sizeof(double) * a[sorted].states);
}
}
for(k = 0; k < tr->rdta->sites; k++)
{
for(j = 0; j < unsortedA[k].states; j++)
fprintf(probsFile, "%f ", unsortedA[k].probs[j]);
fprintf(probsFile, "\n");
fprintf(statesFile, "%c", unsortedA[k].c);
}
fprintf(probsFile, "\n");
fprintf(statesFile, "\n");
}
*counter = *counter + 1;
{
int j;
for(j = 0; j < tr->rdta->sites; j++)
free(unsortedA[j].probs);
for(j = 0; j < tr->cdta->endsite; j++)
free(a[j].probs);
}
free(a);
free(unsortedA);
}
