| 1 | #include "muscle.h" |
|---|
| 2 | #include "msa.h" |
|---|
| 3 | #include "tree.h" |
|---|
| 4 | #include "clust.h" |
|---|
| 5 | #include "clustsetmsa.h" |
|---|
| 6 | #include "distcalc.h" |
|---|
| 7 | |
|---|
| 8 | static void SaveMSADist(const MSA &msa, MSADist &d, const char *FileName) |
|---|
| 9 | { |
|---|
| 10 | FILE *f = fopen(FileName, "w"); |
|---|
| 11 | if (f == 0) |
|---|
| 12 | Quit("Cannot create %s", FileName); |
|---|
| 13 | |
|---|
| 14 | unsigned n = msa.GetSeqCount(); |
|---|
| 15 | for (unsigned i = 0; i < n; ++i) |
|---|
| 16 | { |
|---|
| 17 | fprintf(f, "%10.10s ", msa.GetSeqName(i)); |
|---|
| 18 | for (unsigned j = 0; j < n; ++j) |
|---|
| 19 | fprintf(f, " %9g", d.ComputeDist(msa, i, j)); |
|---|
| 20 | fprintf(f, "\n"); |
|---|
| 21 | } |
|---|
| 22 | fclose(f); |
|---|
| 23 | } |
|---|
| 24 | |
|---|
| 25 | static void TreeFromMSA_NJ(const MSA &msa, Tree &tree, CLUSTER Cluster, |
|---|
| 26 | DISTANCE Distance, const char *SaveFileName) |
|---|
| 27 | { |
|---|
| 28 | MSADist MD(Distance); |
|---|
| 29 | ClustSetMSA Set(msa, MD); |
|---|
| 30 | |
|---|
| 31 | if (SaveFileName != 0) |
|---|
| 32 | SaveMSADist(msa, MD, SaveFileName); |
|---|
| 33 | |
|---|
| 34 | Clust C; |
|---|
| 35 | C.Create(Set, Cluster); |
|---|
| 36 | |
|---|
| 37 | tree.FromClust(C); |
|---|
| 38 | } |
|---|
| 39 | |
|---|
| 40 | static void SaveDC(const DistCalcMSA &DC, const char *FileName) |
|---|
| 41 | { |
|---|
| 42 | FILE *f = fopen(FileName, "w"); |
|---|
| 43 | if (f == 0) |
|---|
| 44 | Quit("Cannot create %s", FileName); |
|---|
| 45 | |
|---|
| 46 | unsigned n = DC.GetCount(); |
|---|
| 47 | fprintf(f, "%u\n", n); |
|---|
| 48 | float *Dist = new float[n]; |
|---|
| 49 | for (unsigned i = 0; i < n; ++i) |
|---|
| 50 | { |
|---|
| 51 | fprintf(f, "%10.10s ", DC.GetName(i)); |
|---|
| 52 | DC.CalcDistRange(i, Dist); |
|---|
| 53 | for (unsigned j = 0; j < i; ++j) |
|---|
| 54 | fprintf(f, " %9g", Dist[j]); |
|---|
| 55 | fprintf(f, "\n"); |
|---|
| 56 | } |
|---|
| 57 | fclose(f); |
|---|
| 58 | } |
|---|
| 59 | |
|---|
| 60 | static void TreeFromMSA_UPGMA(const MSA &msa, Tree &tree, CLUSTER Cluster, |
|---|
| 61 | DISTANCE Distance, const char *SaveFileName) |
|---|
| 62 | { |
|---|
| 63 | LINKAGE Linkage = LINKAGE_Undefined; |
|---|
| 64 | switch (Cluster) |
|---|
| 65 | { |
|---|
| 66 | case CLUSTER_UPGMA: |
|---|
| 67 | Linkage = LINKAGE_Avg; |
|---|
| 68 | break; |
|---|
| 69 | case CLUSTER_UPGMAMin: |
|---|
| 70 | Linkage = LINKAGE_Min; |
|---|
| 71 | break; |
|---|
| 72 | case CLUSTER_UPGMAMax: |
|---|
| 73 | Linkage = LINKAGE_Max; |
|---|
| 74 | break; |
|---|
| 75 | case CLUSTER_UPGMB: |
|---|
| 76 | Linkage = LINKAGE_Biased; |
|---|
| 77 | break; |
|---|
| 78 | default: |
|---|
| 79 | Quit("TreeFromMSA_UPGMA, CLUSTER_%u not supported", Cluster); |
|---|
| 80 | } |
|---|
| 81 | |
|---|
| 82 | DistCalcMSA DC; |
|---|
| 83 | DC.Init(msa, Distance); |
|---|
| 84 | if (SaveFileName != 0) |
|---|
| 85 | SaveDC(DC, SaveFileName); |
|---|
| 86 | UPGMA2(DC, tree, Linkage); |
|---|
| 87 | } |
|---|
| 88 | |
|---|
| 89 | void TreeFromMSA(const MSA &msa, Tree &tree, CLUSTER Cluster, |
|---|
| 90 | DISTANCE Distance, ROOT Root, const char *SaveFileName) |
|---|
| 91 | { |
|---|
| 92 | if (CLUSTER_NeighborJoining == Cluster) |
|---|
| 93 | TreeFromMSA_NJ(msa, tree, Cluster, Distance, SaveFileName); |
|---|
| 94 | else |
|---|
| 95 | TreeFromMSA_UPGMA(msa, tree, Cluster, Distance, SaveFileName); |
|---|
| 96 | FixRoot(tree, Root); |
|---|
| 97 | } |
|---|