b7c2dcc865207dd23298429f93057c9068acef68
angie
  Wed Dec 9 15:48:34 2020 -0800
hgPhyloPlace: Add config option treeChoices: tab-sep file of {protobuf, metadata, source, description} so we can offer the user a choice between GISAID and public sequence trees (possibly different releases for reproducibility).

diff --git src/hg/hgPhyloPlace/treeToAuspiceJson.c src/hg/hgPhyloPlace/treeToAuspiceJson.c
index 8f6030b..78c9355 100644
--- src/hg/hgPhyloPlace/treeToAuspiceJson.c
+++ src/hg/hgPhyloPlace/treeToAuspiceJson.c
@@ -5,94 +5,97 @@
 
 #include "common.h"
 #include "hash.h"
 #include "hui.h"
 #include "jsonWrite.h"
 #include "linefile.h"
 #include "parsimonyProto.h"
 #include "phyloPlace.h"
 #include "phyloTree.h"
 #include "variantProjector.h"
 
 // Globals
 extern char *chrom;
 extern int chromSize;
 
-static void writeAuspiceMeta(FILE *outF, struct slName *subtreeUserSampleIds)
+static void writeAuspiceMeta(FILE *outF, struct slName *subtreeUserSampleIds, char *source)
 /* Write metadata to configure Auspice display. */
 {
 fprintf(outF,
         "\"meta\": { "
         "\"title\": \"Subtree with %s", subtreeUserSampleIds->name);
 struct slName *sln;
 for (sln = subtreeUserSampleIds->next;  sln != NULL;  sln = sln->next)
     fprintf(outF, ", %s", sln->name);
 fputs("\", "
       "\"panels\": [ \"tree\"] , "
       "\"colorings\": [ "
       "  { \"key\": \"pangolin_lineage\", "
       "    \"title\": \"Pangolin lineage\", \"type\": \"categorical\" },"
       "  { \"key\": \"Nextstrain_clade\","
       "    \"scale\": [ [ \"19B\", \"#EC676D\" ], [ \"19A\", \"#F79E43\" ],"
       "        [ \"20A\", \"#B6D77A\" ], [ \"20C\", \"#8FD4ED\" ],"
       "        [ \"20B\", \"#A692C3\" ] ],"
       "    \"title\": \"Nextstrain Clade\", \"type\": \"categorical\" },"
       "  { \"key\": \"GISAID_clade\","
       "    \"scale\": [ [ \"S\", \"#EC676D\" ], [ \"L\", \"#F79E43\" ], [ \"O\", \"#F9D136\" ],"
       "        [ \"V\", \"#FAEA95\" ], [ \"G\", \"#B6D77A\" ], [ \"GH\", \"#8FD4ED\" ],"
       "        [ \"GR\", \"#A692C3\" ] ],"
       "    \"title\": \"GISAID Clade\", \"type\": \"categorical\" },"
-      "  { \"key\": \"userOrOld\", "
-      "    \"scale\": [ [ \"uploaded sample\", \"#CC0000\"] , [ \"GISAID sample\", \"#000000\"] ],"
+      , outF);
+fprintf(outF, "  { \"key\": \"userOrOld\", "
+        "    \"scale\": [ [ \"uploaded sample\", \"#CC0000\"] , [ \"%s\", \"#000000\"] ],"
         "    \"title\": \"Sample type\", \"type\": \"categorical\" }"
-      "  ] , "
+        , source);
+fputs("  ] , "
 //#*** Filters didn't seem to work... maybe something about the new fetch feature, or do I need to spcify in some other way?
 //#***      "\"filters\": [ \"GISAID_clade\", \"region\", \"country\", \"division\", \"author\" ], "
       "\"display_defaults\": { "
       "  \"branch_label\": \"nuc mutations\" "
       "}, "
       , outF);
 fprintf(outF,
         "\"description\": \"Dataset generated by [UShER web interface]"
         "(%shgPhyloPlace) using the "
         "[usher](https://github.com/yatisht/usher/) program.  "
 //#*** TODO: describe input from which tree was generated: user sample, version of tree, etc.
         , hLocalHostCgiBinUrl());
 fputs("If you have metadata you wish to display, you can now drag on a CSV file and it will be "
       "added into this view, [see here]("NEXTSTRAIN_DRAG_DROP_DOC") "
       "for more info.\"} ,"
       , outF);
 }
 
 static void jsonWriteObjectValue(struct jsonWrite *jw, char *name, char *value)
 /* Write an object with one member, "value", set to value, as most Auspice node attributes are
  * formatted. */
 {
 jsonWriteObjectStart(jw, name);
 jsonWriteString(jw, "value", value);
 jsonWriteObjectEnd(jw);
 }
 
 static void jsonWriteLeafNodeAttributes(struct jsonWrite *jw, char *name,
                                         struct sampleMetadata *met, boolean isUserSample,
+                                        char *source,
                                         char **retUserOrOld, char **retNClade, char **retGClade,
                                         char **retLineage)
 /* Write elements of node_attrs for a sample which may be preexisting and in our metadata hash,
  * or may be a new sample from the user.  Set rets for color categories so parent branches can
  * determine their color categories. */
 {
-*retUserOrOld = isUserSample ? "uploaded sample" : "GISAID sample";
+*retUserOrOld = isUserSample ? "uploaded sample" : source;
 jsonWriteObjectValue(jw, "userOrOld", *retUserOrOld);
 if (met && met->date)
     jsonWriteObjectValue(jw, "date", met->date);
 if (met && met->author)
     {
     jsonWriteObjectValue(jw, "author", met->author);
     // Note: Nextstrain adds paper_url and title when available; they also add author and use
     // a uniquified value (e.g. "author": "Wenjie Tan et al" / "value": "Wenjie Tan et al A")
     }
 *retNClade = isUserSample ? "uploaded sample" : (met && met->nClade) ? met->nClade : NULL;
 if (isNotEmpty(*retNClade))
     jsonWriteObjectValue(jw, "Nextstrain_clade", *retNClade);
 *retGClade = isUserSample ? "uploaded sample" : (met && met->gClade) ? met->gClade : NULL;
 if (isNotEmpty(*retGClade))
     jsonWriteObjectValue(jw, "GISAID_clade", *retGClade);
@@ -270,30 +273,31 @@
     jsonWriteObjectEnd(jw);  // mutations
     jsonWriteObjectEnd(jw); // branch_attrs
     }
 }
 
 struct auspiceJsonInfo
 /* Collection of a bunch of things used when writing out auspice JSON for a subtree, so the
  * recursive function doesn't need a dozen args. */
     {
     struct jsonWrite *jw;
     struct slName *subtreeUserSampleIds;  // Subtree node names for user samples (not from big tree)
     struct geneInfo *geneInfoList;        // Transcript seq & alignment for predicting AA change
     struct seqWindow *gSeqWin;            // Reference genome seq for predicting AA change
     struct hash *sampleMetadata;          // Sample metadata for decorating tree
     int nodeNum;                          // For generating sequential node ID (in absence of name)
+    char *source;                         // Source of non-user sequences in tree (GISAID or public)
     };
 
 static int cmpstringp(const void *p1, const void *p2)
 /* strcmp on pointers to strings, as in 'man qsort' but tolerate NULLs */
 {
 char *s1 = *(char * const *)p1;
 char *s2 = *(char * const *)p2;
 if (s1 && s2)
     return strcmp(s1, s2);
 else if (s1 && !s2)
     return 1;
 else if (s2 && !s1)
     return -1;
 return 0;
 }
@@ -360,31 +364,31 @@
         jsonWriteObjectEnd(aji->jw);
         }
     jsonWriteListEnd(aji->jw);
     if (retUserOrOld)
         *retUserOrOld = majorityMaybe(kidUserOrOld, node->numEdges);
     if (retNClade)
         *retNClade = majorityMaybe(kidNClade, node->numEdges);
     if (retGClade)
         *retGClade = majorityMaybe(kidGClade, node->numEdges);
     if (retLineage)
         *retLineage = majorityMaybe(kidLineage, node->numEdges);
     }
 jsonWriteObjectStart(aji->jw, "node_attrs");
 jsonWriteDouble(aji->jw, "div", depth);
 if (node->numEdges == 0)
-    jsonWriteLeafNodeAttributes(aji->jw, name, met, isUserSample,
+    jsonWriteLeafNodeAttributes(aji->jw, name, met, isUserSample, aji->source,
                                 retUserOrOld, retNClade, retGClade, retLineage);
 else if (retUserOrOld && retGClade && retLineage)
     jsonWriteBranchNodeAttributes(aji->jw, *retUserOrOld, *retNClade, *retGClade, *retLineage);
 jsonWriteObjectEnd(aji->jw);
 }
 
 struct phyloTree *phyloTreeNewNode(char *name)
 /* Alloc & return a new node with no children. */
 {
 struct phyloTree *node;
 AllocVar(node);
 AllocVar(node->ident);
 node->ident->name = cloneString(name);
 return node;
 }
@@ -418,47 +422,48 @@
             }
         struct geneInfo *gi;
         AllocVar(gi);
         gi->psl = genePredToPsl((struct genePred *)gp, chromSize, txLen);
         gi->txSeq = newDnaSeq(seq, txLen, gp->name2);
         slAddHead(&geneInfoList, gi);
         }
     lmCleanup(&lm);
     bigBedFileClose(&bbi);
     }
 slReverse(&geneInfoList);
 return geneInfoList;
 }
 
 void treeToAuspiceJson(struct subtreeInfo *sti, char *db, struct dnaSeq *ref,
-                       char *bigGenePredFile, struct hash *sampleMetadata, char *jsonFile)
+                       char *bigGenePredFile, struct hash *sampleMetadata, char *jsonFile,
+                       char *source)
 /* Write JSON for tree in Nextstrain's Augur/Auspice V2 JSON format
  * (https://github.com/nextstrain/augur/blob/master/augur/data/schema-export-v2.json). */
 {
 struct phyloTree *tree = sti->subtree;
 FILE *outF = mustOpen(jsonFile, "w");
 fputs("{ \"version\": \"v2\", ", outF);
-writeAuspiceMeta(outF, sti->subtreeUserSampleIds);
+writeAuspiceMeta(outF, sti->subtreeUserSampleIds, source);
 // The meta part is mostly constant & easier to just write out, but jsonWrite is better for the
 // nested tree structure.
 struct jsonWrite *jw = jsonWriteNew();
 jsonWriteObjectStart(jw, "tree");
 int nodeNum = 10000; // Auspice.us starting node number for newick -> json
 int depth = 0;
 
 // Add an extra root node because otherwise Auspice won't draw branch from big tree root to subtree
 struct phyloTree *root = phyloTreeNewNode("wrapper");
 phyloAddEdge(root, tree);
 tree = root;
 struct geneInfo *geneInfoList = getGeneInfoList(bigGenePredFile, ref);
 struct seqWindow *gSeqWin = chromSeqWindowNew(db, chrom, 0, chromSize);
 struct auspiceJsonInfo aji = { jw, sti->subtreeUserSampleIds, geneInfoList, gSeqWin,
-                               sampleMetadata, nodeNum };
+                               sampleMetadata, nodeNum, source };
 rTreeToAuspiceJson(tree, depth, &aji, NULL, NULL, NULL, NULL);
 chromSeqWindowFree(&gSeqWin);
 jsonWriteObjectEnd(jw);
 fputs(jw->dy->string, outF);
 jsonWriteFree(&jw);
 fputs("}", outF);
 carefulClose(&outF);
 }