1 #include "muscle.h"
2 #include "textfile.h"
3 #include "seqvect.h"
4 #include "distfunc.h"
5 #include "msa.h"
6 #include "tree.h"
7 #include "clust.h"
8 #include "profile.h"
9 #include "clustsetmsa.h"
10 #include "muscle_context.h"
11
Refine()12 void Refine()
13 {
14 MuscleContext *ctx = getMuscleContext();
15
16 SetOutputFileName(ctx->params.g_pstrOutFileName);
17 SetInputFileName(ctx->params.g_pstrInFileName);
18 SetStartTime();
19
20 SetMaxIters(ctx->params.g_uMaxIters);
21 SetSeqWeightMethod(ctx->params.g_SeqWeight1);
22
23 TextFile fileIn(ctx->params.g_pstrInFileName);
24 MSA msa;
25 msa.FromFile(fileIn);
26
27 const unsigned uSeqCount = msa.GetSeqCount();
28 if (0 == uSeqCount)
29 Quit("No sequences in input file");
30
31 ALPHA Alpha = ALPHA_Undefined;
32 switch (ctx->params.g_SeqType)
33 {
34 case SEQTYPE_Auto:
35 Alpha = msa.GuessAlpha();
36 break;
37
38 case SEQTYPE_Protein:
39 Alpha = ALPHA_Amino;
40 break;
41
42 case SEQTYPE_DNA:
43 Alpha = ALPHA_DNA;
44 break;
45
46 case SEQTYPE_RNA:
47 Alpha = ALPHA_RNA;
48 break;
49
50 default:
51 Quit("Invalid SeqType");
52 }
53 SetAlpha(Alpha);
54 msa.FixAlpha();
55
56 SetPPScore();
57 if (ALPHA_DNA == Alpha || ALPHA_RNA == Alpha)
58 SetPPScore(PPSCORE_SPN);
59
60 MSA::SetIdCount(uSeqCount);
61
62 // Initialize sequence ids.
63 // From this point on, ids must somehow propogate from here.
64 for (unsigned uSeqIndex = 0; uSeqIndex < uSeqCount; ++uSeqIndex)
65 msa.SetSeqId(uSeqIndex, uSeqIndex);
66 SetMuscleInputMSA(msa);
67
68 Tree GuideTree;
69 TreeFromMSA(msa, GuideTree, ctx->params.g_Cluster2, ctx->params.g_Distance2, ctx->params.g_Root2);
70 SetMuscleTree(GuideTree);
71
72 if (ctx->params.g_bAnchors)
73 RefineVert(msa, GuideTree, ctx->params.g_uMaxIters);
74 else
75 RefineHoriz(msa, GuideTree, ctx->params.g_uMaxIters, false, false);
76
77 ValidateMuscleIds(msa);
78 ValidateMuscleIds(GuideTree);
79
80 // TextFile fileOut(g_pstrOutFileName, true);
81 // msa.ToFile(fileOut);
82 MuscleOutput(msa);
83 }
84