void main (int argc, char*argv[])
{
int lline=32000, i,j, ch, jeffnode, inode;
char line[32000], mcmcf[96]="o.multidivtime";
FILE *fmcmc;
double t;
puts("Usage:\n\tTreeTimeJeff <MultidivtimeOutputFile>\n");
if(argc>1) strcpy(mcmcf, argv[1]);
fmcmc=gfopen(mcmcf, "r");
/* Read root node number */
for( ; ; ) {
if(fgets(line, lline, fmcmc) == NULL) error2("EOF mcmc file");
if(strstr(line, "Root node number of master tree is")==NULL) continue;
sscanf(line+35, "%d", &j);
com.ns=j/2+1;
break;
}
printf("Tree has %d taxa.\n\n", com.ns);
/* read tree. JeffNode read into [].branch */
for(; ; ) {
ch=fgetc(fmcmc);
if(ch==EOF) error2("EOF treefile");
if(ch=='(')
{ ungetc(ch,fmcmc); break; }
}
ReadTreeN(fmcmc, &i, &j, 2, 0);
OutTreeN(F0,1,0); FPN(F0); FPN(F0);
/* read posterior time estimates */
for(i=0; i<tree.nnode; i++) nodes[i].age=0;
for( ; ; ) {
if(fgets(line, lline, fmcmc) == NULL) error2("EOF mcmc file");
if(strstr(line, "Actual time node")==NULL) continue;
sscanf(line+17, "%d =%lf", &jeffnode, &t);
if(jeffnode<com.ns) {
if(t>0) nodes[inode].age=t;
}
else {
inode=tree.nnode-1+com.ns-jeffnode;
printf("JeffNode %3d ZihengNode %3d time %9.6f\n", jeffnode,inode+1,t);
nodes[inode].age=t;
if(inode==com.ns) break;
if(jeffnode-nodes[inode].branch != 0)
printf(" node number error. ");
}
}
for(i=0; i<tree.nnode; i++)
if(i!=tree.root) nodes[i].branch=nodes[nodes[i].father].age-nodes[i].age;
FPN(F0); OutTreeN(F0,1,1); FPN(F0);
fclose(fmcmc);
exit(0);
}
void TreeDistanceDistribution (FILE* fout)
{
/* This calculates figure 3.7 of Yang (2006).
This reads the file of all trees (such as 7s.all.trees), and calculates the
distribution of partition distance in all pairwise comparisons.
*/
int i,j,ntree, k,*nib, nsame, IBsame[NS], lpart=0;
char treef[64]="5s.all.trees", *partition;
FILE *ftree;
double mPD[NS], PD1[NS]; /* distribution of partition distances */
puts("Tree file name?");
scanf ("%s", treef);
ftree=gfopen (treef,"r");
fscanf (ftree, "%d%d", &com.ns, &ntree);
printf("%2d sequences %2d trees.\n", com.ns, ntree);
i=(com.ns*2-1)*sizeof(struct TREEN);
if((nodes=(struct TREEN*)malloc(i))==NULL) error2("oom");
lpart = (com.ns-1)*com.ns*sizeof(char);
i = ntree*lpart;
printf("\n%d bytes of space requested.\n", i);
partition = (char*)malloc(i);
nib = (int*)malloc(ntree*sizeof(int));
if (partition==NULL || nib==NULL) error2("out of memory");
puts("\ntree #: mean prop of tree pairs with 0 1 2 ... shared bipartitions\n");
fputs("\ntree #: prop of tree pairs with 0 1 2 ... shared bipartitions\n",fout);
for (i=0; i<ntree; i++) {
ReadTreeN (ftree, &j, &k, 0, 1);
nib[i]=tree.nbranch-com.ns;
Tree2Partition(partition+i*lpart);
}
for(k=0; k<com.ns-3; k++) mPD[k]=0;
for (i=0; i<ntree; i++,FPN(fout)) {
for(k=0; k<com.ns-3; k++) PD1[k]=0;
for (j=0; j<ntree; j++) {
if(j==i) continue;
nsame=NSameBranch(partition+i*lpart,partition+j*lpart, nib[i],nib[j],IBsame);
PD1[nsame] ++;
}
for(k=0; k<com.ns-3; k++) PD1[k] /= (ntree-1.);
for(k=0; k<com.ns-3; k++) mPD[k] = (mPD[k]*i+PD1[k])/(i+1.);
printf("%8d (%5.1f%%):", i+1,(i+1.)/ntree*100);
for(k=0; k<com.ns-3; k++) printf(" %7.4f", mPD[k]);
fprintf(fout, "%8d:", i+1); for(k=0; k<com.ns-3; k++) fprintf(fout, " %7.4f", PD1[k]);
printf("%s", (com.ns<8||(i+1)%100==0 ? "\n" : "\r"));
}
free(partition); free(nodes); free(nib); fclose(ftree);
exit(0);
}
void TreeDistances (FILE* fout)
{
/* I think this is broken after i changed the routine Tree2Partition().
*/
int i,j,ntree, k,*nib, parti2B[NS], nsame, IBsame[NS],nIBsame[NS], lpart=0;
char treef[64]="5s.all.trees", *partition;
FILE *ftree;
double psame, mp, vp;
/*
TreeDistanceDistribution(fout);
*/
puts("\nNumber of identical bi-partitions between trees.\nTree file name?");
scanf ("%s", treef);
ftree=gfopen (treef,"r");
fscanf (ftree, "%d%d", &com.ns, &ntree);
printf("%2d sequences %2d trees.\n", com.ns, ntree);
i=(com.ns*2-1)*sizeof(struct TREEN);
if((nodes=(struct TREEN*)malloc(i))==NULL) error2("oom");
if(ntree<2) error2("ntree");
printf ("\n%d species, %d trees\n", com.ns, ntree);
puts("\n\t1: first vs. rest?\n\t2: all pairwise comparisons?\n");
k=2;
scanf("%d", &k);
lpart=(com.ns-1)*com.ns*sizeof(char);
i=(k==1?2:ntree)*lpart;
printf("\n%d bytes of space requested.\n", i);
partition=(char*)malloc(i);
nib=(int*)malloc(ntree*sizeof(int));
if (partition==NULL || nib==NULL) error2("out of memory");
if(k==2) { /* pairwise comparisons */
fputs("Number of identical bi-partitions in pairwise comparisons\n",fout);
for (i=0; i<ntree; i++) {
ReadTreeN (ftree, &j, &k, 0, 1);
nib[i]=tree.nbranch-com.ns;
Tree2Partition(partition+i*lpart);
}
for (i=0; i<ntree; i++,FPN(F0),FPN(fout)) {
printf("%2d (%2d):", i+1,nib[i]);
fprintf(fout,"%2d (%2d):", i+1,nib[i]);
for (j=0; j<i; j++) {
nsame=NSameBranch(partition+i*lpart,partition+j*lpart, nib[i],nib[j],IBsame);
printf(" %2d", nsame);
fprintf(fout," %2d", nsame);
}
}
}
else { /* first vs. others */
ReadTreeN (ftree, &j, &k, 0, 1);
nib[0]=tree.nbranch-com.ns;
if (nib[0]==0) error2("1st tree is a star tree..");
Tree2Partition (partition);
fputs ("Comparing the first tree with the others\nFirst tree:\n",fout);
OutTreeN(fout,0,0); FPN(fout); OutTreeB(fout); FPN(fout);
fputs ("\nInternal branches in the first tree:\n",fout);
FOR(i,nib[0]) {
k=parti2B[i];
fprintf(fout,"%3d (%2d..%-2d): ( ",
i+1,tree.branches[k][0]+1,tree.branches[k][1]+1);
FOR(j,com.ns) if(partition[i*com.ns+j]) fprintf(fout,"%d ",j+1);
fputs(")\n",fout);
}
if(nodes[tree.root].nson<=2)
fputs("\nRooted tree, results may not be correct.\n",fout);
fputs("\nCorrect internal branches compared with the 1st tree:\n",fout);
FOR(k,nib[0]) nIBsame[k]=0;
for (i=1,mp=vp=0; i<ntree; i++,FPN(fout)) {
ReadTreeN (ftree, &j, &k, 0, 1);
nib[1]=tree.nbranch-com.ns;
Tree2Partition(partition+lpart);
nsame=NSameBranch (partition,partition+lpart, nib[0],nib[1],IBsame);
psame=nsame/(double)nib[0];
FOR(k,nib[0]) nIBsame[k]+=IBsame[k];
fprintf(fout,"1 vs. %3d: %4d: ", i+1,nsame);
FOR(k,nib[0]) if(IBsame[k]) fprintf(fout," %2d", k+1);
printf("1 vs. %5d: %6d/%d %10.4f\n", i+1,nsame,nib[0],psame);
vp += square(psame - mp)*(i-1.)/i;
mp=(mp*(i-1.) + psame)/i;
}
vp=(ntree<=2 ? 0 : sqrt(vp/((ntree-1-1)*(ntree-1.))));
fprintf(fout,"\nmean and S.E. of proportion of identical partitions\n");
fprintf(fout,"between the 1st and all the other %d trees ", ntree-1);
fprintf(fout,"(ignore these if not revelant):\n %.4f +- %.4f\n", mp, vp);
fprintf(fout,"\nNumbers of times, out of %d, ", ntree-1);
fprintf(fout,"interior branches of tree 1 are present");
fputs("\n(This may be bootstrap support for nodes in tree 1)\n",fout);
FOR(k,nib[0]) {
i=tree.branches[parti2B[k]][0]+1; j=tree.branches[parti2B[k]][1]+1;
fprintf(fout,"%3d (%2d..%-2d): %6d (%5.1f%%)\n",
k+1,i,j,nIBsame[k],nIBsame[k]*100./(ntree-1.));
}
void LabelClades(FILE *fout)
{
/* This reads in a tree and scan species names to check whether they form a
paraphyletic group and then label the clade.
It assumes that the tree is unrooted, and so goes through two rounds to check
whether the remaining seqs form a monophyletic clade.
*/
FILE *ftree;
int unrooted=1,iclade, sizeclade, mrca, paraphyl, is, imrca, i,j,k, lasts, haslength;
char key[96]="A", treef[64]="/A/F/flu/HA.all.prankcodon.tre", *p,chosen[NS], *endstr="end";
int *anc[NS-1], loc, bitmask, SI=sizeof(int)*8;
int debug;
printf("Tree file name? ");
scanf ("%s", treef);
printf("Treat tree as unrooted (0 no, 1 yes)? ");
scanf ("%d", &unrooted);
ftree = gfopen (treef,"r");
fscanf (ftree, "%d%d", &com.ns, &j);
if(com.ns<=0) error2("need ns in tree file");
debug = (com.ns<20);
i = (com.ns*2-1)*sizeof(struct TREEN);
if((nodes=(struct TREEN*)malloc(i))==NULL) error2("oom");
for(i=0; i<com.ns*2-1; i++) nodes[i].nodeStr = NULL;
for(i=0; i<com.ns-1; i++) {
anc[i] = (int*)malloc((com.ns/SI+1)*sizeof(int));
if(anc[i]==NULL) error2("oom");
}
ReadTreeN(ftree, &haslength, &j, 1, 0);
fclose(ftree);
if(debug) { OutTreeN(F0, 1, PrNodeNum); FPN(F0); }
for(iclade=0; iclade<com.ns-1; iclade++) {
printf("\nString for selecting sequences (followed by non-digit) (end to end)? ");
scanf("%s", key);
if(strcmp(endstr, key) == 0)
break;
for(i=0; i<com.ns; i++)
chosen[i] = '\0';
k = strlen(key);
for(i=0; i<com.ns; i++) {
if( (p=strstr(com.spname[i], key))
&& !isdigit(p[k]) )
chosen[i] = 1;
}
/*
for(i=0; i<com.ns; i++)
if(strstr(com.spname[i], key)) chosen[i] = 1;
*/
/* look for MRCA, going through two rounds, assuming unrooted tree */
for(imrca=0; imrca<1+unrooted; imrca++) {
if(imrca)
for(i=0; i<com.ns; i++) chosen[i] = 1 - chosen[i];
for(i=0,sizeclade=0; i<com.ns; i++)
if(chosen[i]) {
sizeclade ++;
lasts = i;
}
if(sizeclade <= 1 || sizeclade >= com.ns-1) {
puts("unable to form a clade. <2 seqs.");
break;
}
for(i=0; i<com.ns-1; i++) for(j=0; j<com.ns/SI+1; j++)
anc[i][j] = 0;
for(is=0; is<com.ns; is++) {
if(chosen[is]==0) continue;
loc = is/SI; bitmask = 1 << (is%SI);
for(j=nodes[is].father; j!=-1; j=nodes[j].father) {
anc[j-com.ns][loc] |= bitmask;
if(is==lasts) {
for(i=0,k=0; i<com.ns; i++)
if(anc[j-com.ns][i/SI] & (1<<(i%SI)))
k ++;
if(k==sizeclade) {
mrca = j; break;
}
}
}
}
if(imrca==0 && mrca!=tree.root) /* 1st round is enough */
break;
}
if(sizeclade <= 1 || sizeclade >= com.ns-1 || mrca==tree.root) {
printf("Unable to label. Ignored.");
continue;
}
if(debug)
for(is=0; is<com.ns-1; is++) {
printf("\nnode %4d: ", is+com.ns);
for(j=0; j<com.ns; j++) {
loc = j/SI; bitmask = 1 << (j%SI);
printf(" %d", (anc[is][loc] & bitmask) != 0);
}
}
printf("\nClade #%d (%s): %d seqs selected, MRCA is %d\n", iclade+1, key, sizeclade, mrca+1);
for(is=0,paraphyl=0; is<com.ns; is++) {
if(chosen[is] == 0)
for(j=nodes[is].father; j!=-1; j=nodes[j].father)
if(j==mrca) { paraphyl++; break; }
}
if(paraphyl)
printf("\nThis clade is paraphyletic, & includes %d other sequences\n", paraphyl);
nodes[mrca].label = iclade+1;
if(debug) OutTreeN(F0, 1, haslength|PrLabel);
}
for(i=0; i<com.ns-1; i++) free(anc[i]);
OutTreeN(fout, 1, haslength|PrLabel); FPN(fout);
printf("Printed final tree with labels in evolver.out\n");
exit(0);
}
