add APOBEC3G

examples/nice_example/APOBEC3G_primates_ml.dnd_1

@@ -0,0 +1 @@
+((ponPygAY639869:0.107494,(homSapCCDS13984:0.0312308,(gorGorAY639868:0.0361351,(panTroDQ185513:0.0140133,panPanXM_008975100:0.0253136)5:0.0167225)6:0.00381048)7:0.0448956)8:0.0899657,((colGueKC176189:0.0650861,(traFraKC176192:0.040492,(rhiRoxNM_001345905:0.0353039,nasLarKC176190:0.0381885)13:0.0118188)14:0.0415264)15:0.0923819,((allNigKC176173:0.0711026,((cerPetKC176193:0.019077,(cerCepKC176178:0.0204168,(cerWolKC176188:0.0167493,cerNegKC176175:0.0111318)21:0.0104289)22:0.00827238)23:0.0249896,((chlAetAB266486:0.0171719,chlSabXM_007975814:0.00535274)26:0.00712051,(eryPatKC176180:0.0232724,(chlTanJN662547:0.00451739,chlPygJN662545:0.00935603)30:0.00279884)31:0.00280891)32:0.0217719)33:0.0122203)34:0.0265378,((macMulKC176184:0.0216968,(macNemKX583654:0.00812479,macFasAB266488:0.0170656)38:0.00550161)39:0.0617103,(papAnuKC176174:0.0195709,(manLeuNM_001345916:0.0115229,(cerTorKC176182:0.00837242,cerAtyKF020491:0.0196226)44:0.00541445)45:0.0340216)46:0.0460357)47:0.011112)48:0.0433605)49:0.113498);

genere_xml.py

@@ -150,9 +150,8 @@ def nucToAmino(nuc_seq:str):
     return aa
 
 
-def loadAlignment(alignmentFile):
-    '''Loads a FASTA alignment.'''
-
+def loadAlignment(alignmentFile, sites):
+    '''Loads a FASTA alignment with sites (starting at position 0)'''
     alignmentDict = {}
     maxSeqIdLength = 0
     with open(alignmentFile, 'r') as af:
@@ -168,105 +167,106 @@ def loadAlignment(alignmentFile):
                     alignmentDict[seq_id] = aligned_seq
                 else:
                     alignmentDict[seq_id] += aligned_seq
-    return alignmentDict, maxSeqIdLength
 
+    ## Take only assigned sites
+    lenseq=len(alignmentDict[list(alignmentDict.keys())[0]])
+
+    ## Guess is codon sequence or not
+    Msites = max(sites)
+    if Msites>=lenseq/3:
+      isCodon=False # Unsafe:perhaps to few sites to detect non-coding sequence
+    elif Msites<=lenseq:
+      isCodon=True                
+    else:
+      sys.exit("Mismatch lengths between alignment (" + str(lenseq) + ") and sites (" + str(Msites) + ")")
 
-def loadResultsSites(resultsFile, exprcol="'[1]'", nostat_value=-1.0, skipMissingSites=False):
-    '''Loads site results from formula exprcol on columns.'''
+    if not isCodon:
+      lsites = sites
+    else:
+      lsites = [y  for x in sites for y in [3*x,3*x+1,3*x+2]]
+
+    alignmentDict2 = {}
+    for name, seq in alignmentDict.items():
+      alignmentDict2[name]="".join([seq[x] for x in lsites])
+
+    return alignmentDict2, maxSeqIdLength
+
+
+def loadResultsSites(resultsFile, exprcol="'[1]'"):
+    '''Loads site results from formula exprcol on columns.
+    Returns list of results, list of sites (starting with 0)
+    '''
 
     if len(exprcol)>30:
       return
     expr=exprcol[1:-1].replace("[","float(line[")
     expr=expr.replace("]","])")
-    resultsDict = {}
+    resultsList = []
+    siteList =[]
     with open(resultsFile, 'r') as f:
-        i = -1
         for line in f:
             line = line.strip().split()
-            if re.search('^[0-9]+$', line[0]): # results line
+            lp = re.findall('[0-9]+', line[0]) # results line
+            if len(lp)!=0:
                 try:
-                    site = int(line[0])
+                    site = int(lp[0])-1
                     res = float(eval(expr))
                 except ValueError:
                     print(f"Conversion failed in line {line}")
                 else:
-                    if i == -1:
-                        i = site
-                    while i < site and i < MAX_SITE:
-                        # in case there is no statistic for
-                        # a site, give it a default value
-                        if not skipMissingSites:
-                            # print(f'Site {i} missing ({site})')
-                            resultsDict[i] = nostat_value
-                        i += 1
-                    resultsDict[i] = res
-            i += 1
+                    resultsList.append(res)
+                    siteList.append(site)
 
-    resultsList = []
-    for key, item in resultsDict.items():
-        resultsList.append(item)
-    return resultsList
+    return resultsList, siteList
 
 
-def loadResultsBranchSite(resultsFile, nostat_value=-1.0, skipMissingSites=False):
-    '''Loads branch-site results.'''
+def loadResultsBranchSite(resultsFile):
+    '''Loads branch-site results.
+    Returns dict of lists of results, list of sites (starting with 0)
+    '''
 
     col_lists = []
+
     with open(resultsFile, 'r') as f:
       col_headers=f.readline().strip().split()[1:]
       nbcol=len(col_headers)
       col_lists=[[] for i in range(nbcol)]
-
-      numl = -1
+      siteList=[]
+
       for line in f:
         line = line.strip().split()
-        site = int(line[0])
-
-        ## line: ['0', '0.25866598', '0.66856214', '0.19517522', ...]
-        if numl==-1:
-          numl=site
-        if not skipMissingSites:
-          while numl < site and numl < MAX_SITE:
-          # in case there is no statistic for
-          # a site, give it a default value
+        lp = re.findall('[0-9]+', line[0]) # results line
+        if len(lp)!=0:
+          try:
+            siteList.append(int(lp[0])-1)
             for i in range(nbcol):
-              col_lists[i].append(nostat_value)
-            numl += 1
-        else:
-          numl=site
-
-        for i in range(nbcol):
-          col_lists[i].append(line[i+1])
-
-        numl+=1
+              col_lists[i].append(float(line[i+1]))
+          except ValueError:
+            print(f"Conversion failed in line {line}")
 
     d_cols = {}
+
     for i in range(len(col_lists)):
       d_cols[col_headers[i]] = col_lists[i]
-
+      
       ## col_lists: [['0.00885554', '0.25866598', '0.03920189'], ...]
       ## d_cols {'38': ['0.00885554', '0.25866598', '0.03920189'], ...}
 
     d_cols_2 = dict(d_cols)
     nbs=len(d_cols['0'])
-    print(nbs)
     for col_key in d_cols:
       ## For every branch
       if re.search(r'^[0-9]+$', col_key):
-        col_text = ''
-        ## Add each result to the text
-        for i in range(nbs):
-          col_text += f'{d_cols[col_key][i]}, '
-          ## col_text: '0.1248, 0.12381, ...'
-          ## Add brackets for JSON format
-        col_text = '['+col_text[:-2]+']' # [:-2] to remove last space and comma
-        d_cols_2[col_key] = col_text
+        ## col_text: '0.1248, 0.12381, ...'
+        ## Add brackets for JSON format
+        d_cols_2[col_key] = '[' + ",".join([str(d_cols[col_key][i]) for i in range(nbs)]) + "]"
+
         ## d_cols_2: {1: '[0.1248, 0.12381, ...]', ...}
 
     nb_branches = len(col_lists)
     print(nb_branches, 'branches found in results')
 
-    return d_cols_2
+    return d_cols_2, siteList
 
 
 def getColnames(file):
@@ -326,7 +326,7 @@ def createPhyloXML(fam,newick,results):
     ## the first ID being 0.
 
     # Write a sequence of Tree objects to the given file or handle
-    rd = str(random.randint(0,1000))
+    rd = str(random.randint(0,10000))
     Phylo.write([trees], 'tmpfile-'+rd+'.xml', 'phyloxml')
     file = open('tmpfile-'+rd+'.xml', 'r')
     # Copies all objects in a variable and removes the created file
@@ -401,16 +401,18 @@ def count_repl(match):
             leaf.set('speciesLocation', sp)
             if 'seqdefdico' in globals():
                 if cds in seqdefdico:
-                    leaf.set('defintiion', seqdefdico[cds])
+                    leaf.set('definition', seqdefdico[cds])
 
-            ## Add sequence to leaf
+            ## Add sequence to 'leaf
             if lenres==lenseq/3:
                 isCodon="true"
                 leaf.set('dnaAlign', seq_alg) # coding sequence
                 leaf.set('aaAlign', nucToAmino(seq_alg)) # translated sequence
-            else:
-                isCodon="false"
+            elif lenres<=lenseq:
+                isCodon="false"                
                 leaf.set('aaAlign', seq_alg) # raw sequence (any type)
+            else:
+                sys.exit("Mismatch lengths between alignment (" + str(lenseq) + ") and sites (" + str(lenres) + ")")
 
             evrec.append(leaf)
             element.append(evrec)
@@ -423,14 +425,14 @@ def count_repl(match):
                 ## Look for the column with header <branch_id> in the results
                 branch_info = etree.Element('branch_info')
                 branch_info.set('id', branch_id)
-                branch_info.set('results', str(dict_results[branch_id]))
+                branch_info.set('results', str(results[branch_id]))
                 element.append(branch_info)
             except KeyError:
                 ## If there is no matching column is results, add one with negative results
                 print(f'Branch ID {branch_id} not found, adding placeholder')
                 branch_info = etree.Element('branch_info')
                 branch_info.set('id', branch_id)
-                dummy_col = json.loads(dict_results['0'])
+                dummy_col = json.loads(results['0'])
                 dummy_col = [-1.0 for _ in range(len(dummy_col))]
                 col_str = json.dumps(dummy_col)
                 branch_info.set('results', col_str)
@@ -476,20 +478,20 @@ def count_repl(match):
 MAX_SITE = 100000
 sys.setrecursionlimit(15000)
 
-print ("Loading alignment... ")
-loadedAlignment = loadAlignment(args.alignmentFile)
-alignmentDict, maxSeqIdLength = loadedAlignment[0], loadedAlignment[1]
-print ("OK")
-
 print ("Loading results... ")
+sites=[]
 if not args.isBranchsite:
-  results =  loadResultsSites(args.resultsFile, args.exprcol, args.nostat, skipMissingSites=args.skipMissingSites)
+  results, sites = loadResultsSites(args.resultsFile, args.exprcol)
 else:
-  results = loadResultsBranchSite(args.resultsFile)
+  results, sites = loadResultsBranchSite(args.resultsFile)
+
+print("Results read")
+
+print ("Loading alignment... ")
+loadedAlignment = loadAlignment(args.alignmentFile, sites)
+alignmentDict, maxSeqIdLength = loadedAlignment[0], loadedAlignment[1]
+print ("OK")
 
-if args.isBranchsite:
-  dict_results = loadResultsBranchSite(args.resultsFile)
-print("OK")
 
 # Creates empty phyloxml document
 # project = Phyloxml()  # uncomment to have a unique xml file
@@ -517,9 +519,12 @@ def count_repl(match):
     #     newick = tline[0]
     #     fam = ''
     current_branch = -1
-    phyloxmltree = createPhyloXML("",line,results)
-    xmloutputfile.write(phyloxmltree)
-    print("Tree OK")
+    try:
+      phyloxmltree = createPhyloXML("",line,results)
+      xmloutputfile.write(phyloxmltree)
+      print("Tree OK")
+    except ValueError:
+      sys.exit("Mismatch lengths")
 
 treefile.close()
 xmloutputfile.close()

partials/buttons.ejs

@@ -0,0 +1,94 @@
+<!-- Bouton Action -->
+<div id="action" class="dropdown">
+  <button class="btn dropdown-toggle btn-sm btn-outline-secondary" type="button" data-bs-toggle="dropdown" aria-expanded="false" title="Choose action on node">
+    On click
+  </button>
+  <ul class="dropdown-menu" aria-labelledby="action">
+    <li><a id="branch-res" class="dropdown-item" href="#">View Branch Results</a></li>
+    <li><a id="switch" class="dropdown-item" href="#">Switch children</a></li>
+    <li><a id="collapse" class="dropdown-item" href="#">Collapse/Expand</a></li>
+    <li><a id="subtree" class="dropdown-item" href="#">Sub Tree/Upper Tree</a></li>
+  </ul>
+</div>
+<!-- Bouton Expand -->
+<div class="btn-group" role="group" aria-label="">
+  <button id="globalexpand" class="btn btn-sm btn-outline-secondary" type="button" aria-label="Expand" data-toggle="tooltip" data-placement="top" title="Expand the tree (takes time on big trees!)">
+    <span>Expand tree</span>
+  </button>
+</div>
+<!-- Boutons Width / Height -->
+<div class="btn-group" role="group" aria-label="Tree size buttons">
+  <div class="btn-group" role="group" aria-label="Width buttons">
+    <!-- Bouton Reduce width -->
+    <button id="lessWidth" class="btn btn-sm btn-outline-secondary" type="button" aria-label="Less width" data-toggle="tooltip" data-placement="top" title="Reduce tree width">
+      <span>Width -</span>
+      <span class="glyphicon glyphicon-minus" aria-hidden="true"></span>
+    </button>
+    <!-- Bouton Increase width -->
+    <button id="moreWidth" class="btn btn-sm btn-outline-secondary" type="button" aria-label="More width" data-toggle="tooltip" data-placement="top" title="Increase tree width">
+      <span>Width +</span>
+      <span class="glyphicon glyphicon-plus" aria-hidden="true"></span>
+    </button>
+    <!-- Bouton Increase width -->
+    <button id="loga" class="btn btn-sm btn-outline-secondary" type="button" aria-label="Logarithmic scale" data-toggle="tooltip" data-placement="top" title="Logarithmic scale">
+      <span>Logarithmic</span>
+      <span class="glyphicon glyphicon-plus" aria-hidden="true"></span>
+    </button>          
+  </div>
+  <div class="btn-group" role="group" aria-label="Height buttons">
+    <!-- Bouton Reduce height -->
+    <button id="lessHeigth" class="btn btn-sm btn-outline-secondary" type="button" aria-label="Left Align" data-toggle="tooltip" data-placement="top" title="Reduce tree height">
+      <span>Height -</span>
+      <span class="glyphicon glyphicon-minus" aria-hidden="true"></span>
+    </button>
+    <!-- Bouton Increase height -->
+    <button id="moreHeigth" class="btn btn-sm btn-outline-secondary" type="button" aria-label="Left Align" data-toggle="tooltip" data-placement="top" title="Increase tree height">
+      <span>Height +</span>
+      <span class="glyphicon glyphicon-plus" aria-hidden="true"></span>
+    </button>
+  </div>
+</div>
+<!-- Bouton Alignement -->
+<div id="alignment-type" class="dropdown">
+  <button class="btn dropdown-toggle btn-sm btn-outline-secondary" type="button" data-bs-toggle="dropdown" aria-expanded="false" title="Choose alignment type">
+    Alignment type
+  </button>
+  <ul class="dropdown-menu" aria-labelledby="alignment-type">
+    <li><a id="aminoacid-display" class="dropdown-item" href="#">Amino Acids</a></li>
+    <li><a id="codons-display" class="dropdown-item" href="#">Codons</a></li>
+  </ul>
+</div>
+<div class="btn-group" role="group" aria-label="Redundance">
+  <!-- Bouton redondance -->
+  <button id="redondance" class="btn btn-sm btn-outline-secondary" type="button" aria-label="Redundancy" data-toggle="tooltip" data-placement="top" title="Reduce tree height">
+    <span>Redundancy</span>
+  </button>
+</div>
+<!-- Bouton Export -->
+<div id="export" class="dropdown">
+  <button class="btn dropdown-toggle btn-sm btn-outline-secondary" type="button" data-bs-toggle="dropdown" aria-expanded="false" title="Choose Export">
+    Export
+  </button>
+  <ul class="dropdown-menu" aria-labelledby="export">
+    <li><a id="tree" class="dropdown-item" href="#">Tree</a></li>
+    <li><a id="selecp" class="dropdown-item" href="#">Alignment</a></li>
+  </ul>
+</div>
+<!-- Champs Seuils sup. et inf. -->
+<div id="thresholds-group" class="btn-group" role="group" aria-label="">
+  <button class="btn btn-sm btn-outline-secondary">
+    <div class="input-group align-items-center">
+      <label for="ps-threshold-low">Thresholds</label>
+      <input id="ps-threshold-low"  placeholder="PS Threshold" value="0.9" style="width: 5em;"/>
+      <input id="ps-threshold-high" placeholder="PS Threshold" value="1.0" style="width: 5em;"/>
+    </div>
+  </button>
+  <div>
+    <div class="input-group">
+      <input type="radio" class="btn-check" name="selectedThreshold" id="lowerSelected" value="lowerSelected">
+      <label class="btn btn-outline-secondary" for="lowerSelected" style="margin-left: 0.5rem;"><</label>
+      <input type="radio" class="btn-check" name="selectedThreshold" id="upperSelected" value="upperSelected" checked>
+      <label class="btn btn-outline-secondary" for="upperSelected" style="margin-left: -0.25rem;">></label>
+    </div>
+  </div>
+</div>

routes/index.js

@@ -68,12 +68,10 @@ router.post("/upload_files", upload.fields([
   var resultsType = req.body.resultsType;
   var branchSite;
   var logBranchLength = req.body.logBranchLength;
-  var skipMissingSites = req.body.skipMissingSites;
   var isNuc = req.body.isNuc;
 
   isNuc = (isNuc != undefined ? true : false);
   logBranchLength = (logBranchLength != undefined ? true : false);
-  skipMissingSites = (skipMissingSites != undefined ? true : false);
   branchSite = (resultsType == 'branchSiteMode' ? true : false);
 
   // Generate XML file from data
@@ -85,8 +83,7 @@ router.post("/upload_files", upload.fields([
       +' -o '+xml_dir+fname_xml
       +' -e "\''+statcol+'\'"'
       +' -n '+nostat
-      +(branchSite?' -b ':'')
-      +(skipMissingSites?' --skipmissing ':''),
+      +(branchSite?' -b ':''),
   (error, stdout, stderr) => {
     if (error) {
       console.log(`error: ${error.message}`);