20190612 - Meeting

[javirudolph]

Meeting with Stuart

He is suggesting a tradeoff between making volatiles and the leaf size, in the case of males. Repro unit and some clusters. Total amount of carbon produced by those clusters, compared to all the carbon that is produced. In the case of males, interesting to see negative correlation between the vocs and several life history traits.

• Interesting to see that a lot of clusters are negatively correlated with reproductive variable.
• Look for some that would have positive correlations with the reproductive/gam traits.

Missing the correlation matrix of the traits for females.

Look at the correlation matrices for males and females side by side, or subtract one from the other to see the change between them.

Regarding the talk

• Introduce people to mosses and how you did this experiment
• Start with the data from Todd's paper
  • Bugs prefer females and that has a great benefit for the plant, increasing fertilization rates.
• Can dimorphism evolve or is it constrained by genetic covariances?
• We see dimorphism, so there should be a cost for males to make these volatiles. You see this with the tradeoff between making these compounds and the production of sex structures. Need to do fitness stuff first to show this, but correlations work for now.
  • Are these clusters sexually dimorphic??
  • Use the boxplot of males vs females for the VOCs clusters and point out the ones where you see sexual dimorphism.
  • What explains the dimorphism: know that they are beneficial for females, which explains why they are costly for males.
  • If they are costly for males, you would expect a negative correlation between volatile production and male reproduction (the repro variable). And point this out in the figure by highlighting the volatile clusters that show this dimorphism from the anovas or manovas.
• Where are these compounds produced? Talk about the negative correlation with leaves.
• Differences worth mentioning: females having negative correlation between reproduction and the growth measurements.
• Highlight particular matrices that are important
  • volatiles: free to evolve, because there is no strong correlation between males and females.
  • selection in males is not going to have a strong response in females.

[Kollarlm]

This is Stuart's feedback after the talk...

Here is the review that I mentioned -- https://onlinelibrary.wiley.com/doi/full/10.1111/jeb.12188

I think starting out your talk with a bit about sexual dimorphism – most significant polymorphism within many populations, often in traits related to reproduction, but curiously in other traits seemingly less connected to reproduction too. The evolution of sexual dimorphism is constrained by the fact that one genome must build two distinct phenotypes (female and male), a constraint partially described by the cross sex correlation. In a quantitative genetics framework, the cross sex correlations form the diagonals of a cross sex variance covariance matrix, called B. However, the off-diagonal elements of B describe how the evolution of dimorphism potentially can be enhanced by trait covariation, in particular if B is asymmetric.

Here we examine dimorphism in moss VOCs. These are known to be important for mating… etc…

I would think about showing data on additive vs environmental variance for VOC production

A statistical test comparing G male vs G female

And also comparing the symmetry in the B matrix – is above the diagonal the same as below?

[javirudolph]

VOC with clones

I created a new .md file here. It is in the Rmds folder.

It creates a new data frame .RDS file with the VOC data, scaled and transformed with the clones info.

However, when doing the clustering, we get a different number of clusters (19). I don't trust this or the correlation matrix from it, because we are considering each clone as an independent sample, but we know this isn't true since they are related by their genotype. Will have to look into this more.

• [ ] Look into correlations with non-independent samples?
• [ ] Alternatives to clustering based on the correlation coefficients?

Overall, I think it's important to include the fact that these are clones and therefore not independent samples?

[Kollarlm]

Ok. So getting the clusters is not necessarily possible? I think I understand what you are saying.

Leslie

Leslie M. Kollar

Ph.D. Candidate

Department of Biology

University of Florida

Twitter: @Kollar_Genetics

Pronouns: She/her/hers

---

From: Javiera Rudolph notifications@github.com Sent: Monday, June 17, 2019 11:11:27 AM To: javirudolph/mossmat Cc: Kollar,Leslie M; Comment Subject: Re: [javirudolph/mossmat] 20190612 - Meeting (#9)

VOC with clones

I created a new .md file herehttps://urldefense.proofpoint.com/v2/url?u=https-3A__github.com_javirudolph_mossmat_blob_master_Rmds_VOC-5Fcleanup-5Fw-5Fclones.md&d=DwMCaQ&c=sJ6xIWYx-zLMB3EPkvcnVg&r=XBGf66T2ZPyLIsd9UZy7Vw&m=mBd9YDzouqLo0eTN-nYarrm_Eaclgcp_HAfQOUoPKp4&s=iuDqSnIrvC-kFUpnKem5eIypSBO285JL_0HX4R0OAGU&e=. It is in the Rmds folder.

It creates a new data frame .RDS file with the VOC data, scaled and transformed with the clones info.

However, when doing the clustering, we get a different number of clusters (19). I don't trust this or the correlation matrix from it, because we are considering each clone as an independent sample, but we know this isn't true since they are related by their genotype. Will have to look into this more.

• Look into correlations with non-independent samples?
• Alternatives to clustering based on the correlation coefficients?

Overall, I think it's important to include the fact that these are clones and therefore not independent samples?

— You are receiving this because you commented. Reply to this email directly, view it on GitHubhttps://urldefense.proofpoint.com/v2/url?u=https-3A__github.com_javirudolph_mossmat_issues_9-3Femail-5Fsource-3Dnotifications-26email-5Ftoken-3DACP27O5EFTHGKQNUZJENDXDP26SR7A5CNFSM4HX4PQQ2YY3PNVWWK3TUL52HS4DFVREXG43VMVBW63LNMVXHJKTDN5WW2ZLOORPWSZGODX3PRQQ-23issuecomment-2D502724802&d=DwMCaQ&c=sJ6xIWYx-zLMB3EPkvcnVg&r=XBGf66T2ZPyLIsd9UZy7Vw&m=mBd9YDzouqLo0eTN-nYarrm_Eaclgcp_HAfQOUoPKp4&s=wjKfpzxHLTcA20hSrbY36k8zUiQPAnLOh6YEyyfIwPg&e=, or mute the threadhttps://urldefense.proofpoint.com/v2/url?u=https-3A__github.com_notifications_unsubscribe-2Dauth_ACP27O2IYJFOYZFMPONQJ7TP26SR7ANCNFSM4HX4PQQQ&d=DwMCaQ&c=sJ6xIWYx-zLMB3EPkvcnVg&r=XBGf66T2ZPyLIsd9UZy7Vw&m=mBd9YDzouqLo0eTN-nYarrm_Eaclgcp_HAfQOUoPKp4&s=6sqzePVyR3tDq9MXjZCLujFW7e9j-74NVccN_ndOdcc&e=.

[javirudolph]

We can get the clusters! Just not using exactly the same method. So, I need to look into it and will probably not happen today.