How to normalize distance

[guojun wu]

Hi all,

I am reading the paper "Mother-to-Infant Microbial Transmission from Different Body Sites Shapes the Developing Infant Gut Microbiome", which used MetaPhlAn and StrainPhlAn.

In this paper,"Comparison among different trees was performed by normalizing distances in each tree by its median value. Finally, strain distance for any pair of samples was defined as the normalized phylogenetic distance on the corresponding tree". The authors (including StrainPhlAn authors) normalized distances in each tree by its median value. I am confused how to do it based on the RAxML tree file from StrainPhlAn. Does anyone know more details?  Thank you!

[Francesco Asnicar]

Hi,

The idea is that since each phylogenetic tree will have a different height, so you need to define a way to normalize the tree to set a threshold that will hold for any phylogeny. From the tree computed by RAxML, you can compute the median of the pairwise phylogenetic distances within the tree and use that as a normalization factor. That was then used to define the strain distance.

I hope this helps.

Many thanks,

Francesco

--
You received this message because you are subscribed to the Google Groups "MetaPhlAn-users" group.
To unsubscribe from this group and stop receiving emails from it, send an email to metaphlan-use...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/metaphlan-users/d9e631f5-0a18-4326-81b7-4f8253551c93%40googlegroups.com.

[guojun wu]

Hi Francesco,

Thank you for your reply. I am sorry I am not familiar with median normalization. When you say "as a normalization factor", do you mean (pairwise phylogenetic distances)/median(distance)?

On Sunday, October 6, 2019 at 5:45:37 AM UTC-4, Francesco Asnicar wrote:

Hi,

The idea is that since each phylogenetic tree will have a different height, so you need to define a way to normalize the tree to set a threshold that will hold for any phylogeny. From the tree computed by RAxML, you can compute the median of the pairwise phylogenetic distances within the tree and use that as a normalization factor. That was then used to define the strain distance.

I hope this helps.

Many thanks,

Francesco

On Sun, Sep 22, 2019 at 11:12 PM guojun wu <wuguo...@gmail.com> wrote:

Hi all,

I am reading the paper "Mother-to-Infant Microbial Transmission from Different Body Sites Shapes the Developing Infant Gut Microbiome", which used MetaPhlAn and StrainPhlAn.

In this paper,"Comparison among different trees was performed by normalizing distances in each tree by its median value. Finally, strain distance for any pair of samples was defined as the normalized phylogenetic distance on the corresponding tree". The authors (including StrainPhlAn authors) normalized distances in each tree by its median value. I am confused how to do it based on the RAxML tree file from StrainPhlAn. Does anyone know more details?  Thank you!

--
You received this message because you are subscribed to the Google Groups "MetaPhlAn-users" group.

To unsubscribe from this group and stop receiving emails from it, send an email to metaphl...@googlegroups.com.

[Francesco Asnicar]

Hi, yes, that's right. Each pairwise phylogenetic distance divided by the median of all pairwise phylogenetic distances.

Best,
Francesco

To unsubscribe from this group and stop receiving emails from it, send an email to metaphlan-use...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/metaphlan-users/071ec54d-e130-4646-ab2f-920747245190%40googlegroups.com.