WIP: Try best rooting

[j23414]

Description of proposed changes

This draft PR differs in that it is mostly formatted as a continuation of an ongoing discussion about fixed root versus allowing TreeTime to infer the root of Lassa virus phylogenetic trees. The discussion was initially started by @JoiRichi and @j23414 but is open to input from other contributors. This PR acts as a documented code exploration of the topics discussed, providing a practical examination of the concepts and ideas raised during the conversation.

build	Compare Before (left) and After (right)	before clock rate (6e-4)	after clock rate (let treetime estimate)
l	https://next.nextstrain.org/lassa/l:staging/lassa/trials/rerooting/lassa/l	1.55e-3	5.21e-7
s	https://next.nextstrain.org/lassa/s:staging/lassa/trials/rerooting/lassa/s	1.29e-3	1.25e-5

Related issue(s)

• Compare TreeTime inferred rooting to set root #44

Checklist

• Checks pass

[j23414]

Comparison of Tree Topologies: Fixed Root vs. the default Treetime "Best" Root

The change in root determination method from fixed to "best" has varying impacts on the L and S segments:

• L Segment: The transition results in a mostly consistent tree topology compared to the Fixed Root.
• S Segment: The change leads to a more pronounced difference in tree structure.

Visual Comparison

The attached image illustrates the contrast:

• Left: Fixed root S segment tree
• Right: "Best" root S segment tree (as determined by TreeTime)

Refinement Command

The following augur refine command was used to generate the "best" root tree:

augur refine  \
--tree results/s/tree_raw.nwk \
--alignment results/s/aligned.fasta  \
--metadata data/s/metadata.tsv \
--metadata-id-columns accession   \
--output-tree results/s/tree.nwk \
--output-node-data results/s/branch_lengths.json  \
--timetree \
--coalescent opt  \
--date-confidence \
--date-inference marginal \
--root best         #<= pick rooting method here

I'll run a mid_point root to compare.

@JoiRichi, I'm going to assume we're looking for a S tree topology that is mostly consistent with L segment trees? Or at least that the L segment tree may more closely match the evolutionary history of the virus?

[j23414]

I've uploaded a mid_point S tree at

• https://next.nextstrain.org/staging/lassa/s/midpoint

Let me know if this looks better than the fixed_root or feel free to suggest other rooting methods:

augur refine -h
# ...
  --root ROOT [ROOT ...]
                        rooting mechanism ('best', least-squares', 'min_dev',
                        'oldest', 'mid_point') OR node to root by OR two nodes
                        indicating a monophyletic group to root by. Run
                        treetime -h for definitions of rooting methods.
                        (default: best)
# ...

[JoiRichi]

The literature has shown that the evolution of LASV is complex. We may not expect same tree topology because: Both proteins encoded by the S segment are immunogenic and may therefore be subjected to immune pressure. If the L segment isn’t, then we may not expect same evolution pattern. There are evidences of reassortment and recombination across segments. This almost guarantees different evolution patterns across the segments. From paper below: “ Interestingly, although the length of S segment is less than half of the L segment, the proportion of LASVs with S recombinants is significantly higher than that with L recombinants. These results suggest that HR may be a feature of LASV, which can be set by natural selection to produce beneficial or eliminate harmful mutations for the virus, so it plays a role in LASV evolution during the outbreak of LF.” https://www.frontiersin.org/journals/microbiology/articles/10.3389/fmicb.2024.1411537/full So my suggestion may be to treat the segments differently.

…

On Fri 29. Nov 2024 at 16:49, Jennifer Chang ***@***.***> wrote: Comparison of Tree Topologies: Fixed Root vs. "Best" Root The change in root determination method from fixed to "best" has varying impacts on the L and S segments: - *L Segment*: The transition results in a mostly consistent tree topology compared to the Fixed Root. - *S Segment*: The change leads to a more pronounced difference in tree structure. Visual Comparison The attached image illustrates the contrast: - Left: Fixed root S segment tree - Right: "Best" root S segment tree (as determined by TreeTime) Screenshot.2024-11-29.at.7.39.48.AM.png (view on web) <https://github.com/user-attachments/assets/8e0d321b-7fce-44a4-a15a-9f9f4a6b1625> Refinement Command The following augur refine command was used to generate the "best" root tree: augur refine \ --tree results/s/tree_raw.nwk \ --alignment results/s/aligned.fasta \ --metadata data/s/metadata.tsv \ --metadata-id-columns accession \ --output-tree results/s/tree.nwk \ --output-node-data results/s/branch_lengths.json \ --timetree \ --coalescent opt \ --date-confidence \ --date-inference marginal \ --root best #<= pick rooting method here I'll run a mid_point root to compare. @JoiRichi <https://github.com/JoiRichi>, I'm going to assume we're looking for a S tree topology that is mostly consistent with L segment trees? Or at least that the L segment tree may more closely match the evolutionary history of the virus? — Reply to this email directly, view it on GitHub <#46 (comment)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/AOVPUJTVHCP6U54JG6NDNHD2DCEJ7AVCNFSM6AAAAABSVWSPIOVHI2DSMVQWIX3LMV43OSLTON2WKQ3PNVWWK3TUHMZDKMBYGA3DIOBXHE> . You are receiving this because you were mentioned.Message ID: ***@***.***>