`read.mega` / `read.beast` returns incorrect tip node data when reading nexus file with non-consecutive translation table keys

[Fan-iX]

If the translation table keys of the nexus file are not consecutive integers (which is possible for a MEGA timetree analysis output), read.mega / read.beast will return a incorrect @data table:

test.nex

#NEXUS

Begin Taxa;
	Dimensions ntax=4;
	TaxLabels
		A
		B
		C
		D
		;
End;

Begin Trees;
	Translate
		1 A,
		2 B,
		4 C,
		5 D
		;
tree TREE1 = [&R] (((1[&rate=NA,branch_length=3.05210780043533E-003]:2.82146551149507E-003,2[&rate=NA,branch_length=3.11763249917918E-003]:2.82146551149507E-003)[&rate=NA,branch_length=3.27365421500745E-002,reltime=2.82146551149507E-003,reltime_stderr=3.48412420431308E-004,reltime_95%_CI={2.22004192087475E-003,3.58581860896548E-003},data_coverage=99%]:2.99429893613024E-002,4[&rate=NA,branch_length=2.83944102398042E-002]:3.27644548727974E-002)[&rate=NA,branch_length=4.76537203824402E-002,reltime=3.27644548727974E-002,reltime_stderr=2.55675381204422E-003,reltime_95%_CI={2.81342298521227E-002,3.81567047953361E-002},data_coverage=99%]:4.89568013527124E-002,5[&rate=NA,branch_length=8.39559068047214E-002]:8.17212562255099E-002)[&rate=NA,branch_length=1.28935725106037E-002,reltime=8.17212562255099E-002,reltime_stderr=4.03743281925163E-003,reltime_95%_CI={7.41901341411784E-002,9.00168707926448E-002},data_coverage=99%];
End;

---

> read.beast("text.nex")
'treedata' S4 object that stored information of
	'text.nex'.

...@ phylo:

Phylogenetic tree with 4 tips and 3 internal nodes.

Tip labels:
  A, B, C, D

Rooted; includes branch length(s).

with the following features available:
  'branch_length', 'data_coverage', 'rate', 'reltime', 'reltime_0.95_CI', 'reltime_stderr'.

# The associated data tibble abstraction: 8 × 9
# The 'node', 'label' and 'isTip' are from the phylo tree.
   node label isTip branch_length       data_coverage rate  reltime             reltime_0.95_CI reltime_stderr      
  <int> <chr> <lgl> <chr>               <chr>         <chr> <chr>               <list>          <chr>               
1     1 A     TRUE  0.00305210780043533 NA            NA    NA                  <chr [1]>       NA                  
2     2 B     TRUE  0.00311763249917918 NA            NA    NA                  <chr [1]>       NA                  
3     3 C     TRUE  NA                  NA            NA    NA                  <NULL>          NA                  
4     4 D     TRUE  0.0283944102398042  NA            NA    NA                  <chr [1]>       NA                  
5     5 NA    FALSE 0.0839559068047214  NA            NA    NA                  <chr [1]>       NA                  
6     5 NA    FALSE 0.0128935725106037  99%           NA    0.0817212562255099  <dbl [2]>       0.00403743281925163 
7     6 NA    FALSE 0.0476537203824402  99%           NA    0.0327644548727974  <dbl [2]>       0.00255675381204422 
8     7 NA    FALSE 0.0327365421500745  99%           NA    0.00282146551149507 <dbl [2]>       0.000348412420431308

Note the node 3 (labeled C) has no associated data, which is not the case in the nexus file. Also, node 5 has two different entries.

The expected result will be:

   node label isTip branch_length       data_coverage rate  reltime             reltime_0.95_CI reltime_stderr      
  <int> <chr> <lgl> <chr>               <chr>         <chr> <chr>               <list>          <chr>               
1     1 A     TRUE  0.00305210780043533 NA            NA    NA                  <chr [1]>       NA                  
2     2 B     TRUE  0.00311763249917918 NA            NA    NA                  <chr [1]>       NA                  
3     3 C     TRUE  0.0283944102398042  NA            NA    NA                  <chr [1]>       NA                  
4     4 D     TRUE  0.0839559068047214  NA            NA    NA                  <chr [1]>       NA                  
5     5 NA    FALSE 0.0128935725106037  99%           NA    0.0817212562255099  <dbl [2]>       0.00403743281925163 
6     6 NA    FALSE 0.0476537203824402  99%           NA    0.0327644548727974  <dbl [2]>       0.00255675381204422 
7     7 NA    FALSE 0.0327365421500745  99%           NA    0.00282146551149507 <dbl [2]>       0.000348412420431308

This issue arises after a recent fix of the ape package, which relaxed the restriction of the translation table when reading nexus files. Before that fix, translation table containing non-consecutive keys are not supported. (see ape#133 )

[Fan-iX]

The issue is caused by a misinterpretation of tree nodes in read.stats_beast_internal

treeio/R/beast.R

Lines 197 to 201 in 3f53eec

	if (is_translated == TRUE) {
	label2 <- c(phylo$tip.label, root:getNodeNum(phylo))
	} else {
	label2 <- as.character(1:getNodeNum(phylo))
	}

If phylo$tip.label are not consecutive integers, label2 will no longer cover the range of tip nodes and lead to a shift.

Currently ape uses different parsing strategies for nexus files with or without consecutive translation key, and by now label2 <- c(phylo$tip.label, root:getNodeNum(phylo)) shall be used only if is_translated == TRUE and the nexus files contains consecutive translation key.

[brj1]

This should be fixed in the second commit of my PR #130 for read.beast.

[Fan-iX]

@brj1 I tried the PR, it gives exactly the same output.

I'm afraid it doesn't fix this problem.

[Fan-iX]

The latest version of ape(>=5.8-1.3) has changed the parsing strategies for nexus files, now simple

-  if (is_translated == TRUE) { 
-      label2 <- c(phylo$tip.label, root:getNodeNum(phylo)) 
- } else { 
     label2 <- as.character(1:getNodeNum(phylo)) 
- } 

seems to make it.