missing tips in newick tree

[Djeppschmidt]

Hello,

I'm really appreciative of the newick format that you recently introduced!

I think this is a bug in building the tree. As I'm working with the newick file, it appears the newick tree is missing internal nodes; rather about half the nodes are labeled with the names that should actually be tips on the tree. For example, I ran rabbitTclust to cluster all salmonella in the NCBI pathogen database (~500k isolates) using the following code:

clust-mst -d 0.001 -l -i fasta_input.txt --newick-tree -o sal.mst.clust.0001

I generate a tree with ~270k tips, and ~238k nodes (it should have ~500k tips).

I ran a tiny version of this with 8 isolates, which produced 3 tips, and 5 internal nodes:

(((/isilon/NCBI/SRAassemblies/skesa_contigs/SRR863221_contigs_skesa.fasta:0.000794,(/isilon/NCBI/SRAassemblies/skesa_contigs/SRR863395_contigs_skesa.fasta:0.016157)/isilon/NCBI/SRAassemblies/skesa_contigs/SRR900926_contigs_skesa.fasta:0.000969,(/isilon/NCBI/SRAassemblies/skesa_contigs/SRR863393_contigs_skesa.fasta:0.001294)/isilon/NCBI/SRAassemblies/skesa_contigs/SRR863392_contigs_skesa.fasta:0.013981)/isilon/NCBI/SRAassemblies/skesa_contigs/SRR863223_contigs_skesa.fasta:0.000000)/isilon/NCBI/SRAassemblies/skesa_contigs/SRR863224_contigs_skesa.fasta:0.020389)/isilon/NCBI/SRAassemblies/skesa_contigs/SRR863396_contigs_skesa.fasta;

This makes it impossible to filter the tree by tips because half the isolates are actually node labels, when I believe they should be tip labels.

I'm curious if anyone else is experiencing this issue? Or maybe I'm missing something?

Thanks for you help, Dietrich

[XiaomingXu1995]

Hi Dietrich,

Thanks for your issue!

I am currently on a business trip this month. I will check it and let you know when I have progress.

Best, Xiaoming

[avw-adifranco]

Hi Xiaoming,

I got the same issue as Dietrich as I was hoping the tree to output each cluster as a leave. Instead, most clusters are actually present as named internal nodes.

Did you had some times to look into it ? I believe the reason could come from the presence of 0 in the distance matrix as some sequence could be considered as subsets of the other. Maybe replacing those 0 by a really small distance value could produce what Dietrich and I would expect.

If you could pinpoint in your code where the newick tree is done, I could look more into it.

Best, Arnaud

[XiaomingXu1995]

Apologies for the delayed response.

The Newick Tree in RabbitTClust represents the output format of the Minimum Spanning Tree generated in clust-mst. Unfortunately, it is not possible to designate all genome nodes as leaf nodes, as the connections of the edges in the Minimum Spanning Tree are dependent on internal nodes.

Best, Xiaoming