You can use the msa package and a few others for the Bayesian and ML analyses you would usually see in softwares like MEGA. Here's an rbubs blog that goes through the basic process: rpubs.com/mvillalobos/L01_Phylogeny

You can just use gsub (the find and replace function in R) with a regex to remove whatever you want. Like this: your_data$your_column <- gsub("\\.", "", your_data$your_column)

absolutely! So in the code chunk where it says: dat.tab <- data%>% group_by(Grouping_column_ID)%>% count(value)%>% mutate(percent = round(n/sum(n)*100, 1)) You would change the count() function instead to summarize() function and add the new column name avg (or whatever you want to call it) and the mean() function for you numeric values (lets call them value here also as it should be the output of melt()). It would then look like this: dat.tab <- data%>% group_by(Grouping_column_ID)%>% summarize(avg = mean(value)) With this, you will get the mean for each grouping ID, however you cannot get a percentage in this case because you do not have the sum of the counts but rather the mean of the counts. What you could do, for comparative purposes would be to add maybe a min and max value, like so: dat.tab <- data%>% group_by(Grouping_column_ID)%>% summarize(avg = mean(value), min = min(value), max = max(value)) For these numeric value columns grouped by ID, you could also use great visualization functions from the package ggstatsplot to plot them into boxplots/violin plots with their associated pairwise and groupwise statistical tests library(ggstatsplot) ggbetweenstats(data=data, x = Grouping_coumn_ID, y = value)

Are you using R or Rstudios? Are you on a Mac, Windows, or Linux system?

My immediate guess would be that you're using a Mac and you need to install XQuarts (www.xquartz.org/)

@@RJG_Ecology I am using Windows and R Studio.

@@stephhogan6070So when you run the plot does it show up in the bottom right plot window at all? Or immediately opens an X11 (external) window by itself? Also when you say "I cannot get it to zoom out" what do you mean?

@@RJG_Ecology No, it does not show up in the bottom right plot window. It opens by itself in a new window. When it opens I am unable to see the strongly agree response at the bottom.

In SMART, the initial step is to install the R plugin (on SMART Desktop), which you can then access through the Queries tab. Once you have the plugin, you can get this script from here: github.com/RJGrayEcology/smartR/blob/main/Regression_plots.R Then you add the script in the R plugin on SMART Desktop and run. If you are using the SVW data model, then the script does not need to be modified. If you are using a different data model, you may have to modify the script to represent the names of your data fields.

@@RJG_Ecology many thank you for your reply to me, first of all I would like to tell you that my English is poor some time it might hard to understand my message. continue our point, I already install R plugin and added the script to the R queries, but my issue is my data model is different. So, now II need to know your queries step and see your queries column names for my understanding to edit the R code/script in my data model and queries. Many thanks again.

@@haivvemvang7731 If you have a different data model, then all of the fields (columns) I include in this video should still exist, they just have different names. So you need to query those same fields in SMART, but use the names that exist in your own data model. Also, you might want to make a linear model for things other than traps, camps, and offenders. In which case you can just pick any three data types in your own SMART query and run those instead. If you try it and run into any errors, let me know and I will try to help

@@RJG_Ecology Many thanks, I will try to do it and let you know again.

@@RJG_Ecology could I share my SMART database with you and then can you revise the script and revert to me?

Can you elaborate a bit more on what you want to do? In the meantime here is the ggtree documentation, it might have what you're looking for. 4va.github.io/biodatasci/r-ggtree.html

Hey Steve, no I am not. I was an independent researcher in Belize at this time. Regarding coral snakes in Belize, there are two confirmed species at the moment -- Micrurus hippocrepis and Micrurus diastema. I assume what you saw was the latter; its common name is the "variable coral snake" since its morphology is extremely variable from individual to individual. Regarding your phylogeny project, that sounds like quite the undertaking, NCBI also has something of this nature right now you should check out if you haven't already, called "LifeMap". It's basically a massive interactive phylogency of all known life in a fixed branch tree: lifemap-ncbi.univ-lyon1.fr/

Tons of phylogenetics R packages. For tree building and visualization I would say phylotools, phytools, ape, and ggtree package are the most helpful. RevGadgets package has a mix of everything. see their paper here: besjournals.onlinelibrary.wiley.com/doi/epdf/10.1111/2041-210X.13750

Hey Ugur, the reason for this error in R is that you have not opened the function library. In this case, the function library is "DECIPHER". Make sure you have installed DECIPHER using: if (!requireNamespace("BiocManager", quietly = TRUE)) install.packages("BiocManager") BiocManager::install("DECIPHER") and then open it using: library(DECIPHER)

@@RJG_Ecology Thank you for replying Russ. But I got new error like this: in f(p.profile[, anchors[2, n - 1]:anchors[1, n], drop = FALSE], : Alignment larger (9,174,518,227) than the maximum allowable size (2,147,483,647). How can I fix it?

@@uguremre3287 The maximum allowable size for alignments with DECIPHER alignseqs() is 2,147,483,647. Therefore anything larger will need to use a different alignment function such as FindSynteny() followed by AlignSynteny().

@@RJG_Ecology I tried to run from AlignSynteny() but I couldn't figure out it:(

Error in AlignSynteny(apricot) : synteny must be an object of class 'Synteny'

Hi Pamela, Yes actually! The taxonomizr package (see tutorial here: cran.r-project.org/web/packages/taxonomizr/readme/README.html) has a two step process for this purpose with functions "accessionToTaxa", which convert accession numbers to taxonomic IDs, and then "getTaxonomy" convert taxonomic IDs to taxonomy. They have examples of how to do so in the link. Let me know if you run into any issues!

What line of code are you getting this error and with what data?

Just using defaults for the tutorial, the packages have multiple different methods that can be applied, I'm not sure if this one has Bayesian inference or not, but I agree that or MLE would probably be optimal!

Does anyone know how to build the tree using Maximum Likelihood instead of neighbor-joining?

With tree data do you mean ".tre" files? You can just read those in with the read.tree function and combine them with merge_tree if they have common variables

@@RJG_Ecology thank you sir for reply. Actually my teacher told me to calculate phylogenetic diversity from tree data he send me i can use R but it makes me more confuse since last week i m trying did not find anyway how to do it. If you can guide me about phylogenetic diversity would be very appreciated. Thank you sir.

What lines of code are you running when you get the error?

@@RJG_Ecology temp <- as.data.frame(as.matrix(D)) table.paint(temp, cleg=1, clabel.row=.4, clabel.col=.4)+ scale_color_viridis()

@@Gayensubrata89 it looks like the ade4 "table.paint" function has updated and removed the argument "cleg", just remove that and it should work. i.e. table.paint(temp, clabel.row=.4, clabel.col=.4)+ scale_color_viridis()

@@RJG_Ecology No it is not working. Still having the error- Error in gray(valgris[numclass]) : invalid gray level, must be in [0,1].

@@Gayensubrata89 Please show me the code you ran to get that error.

Link to the code and data is in the description already

The link is working fine on my end. Check your browser and firewall settings, could also be connection. Can you access github by itself? github.com/

That would be something you could do with species distribution models but that would entirely depend on the species of tree and the type of ecosystem. I would look into the packages sdm, biomod2, and ENMtools.

André, would you mind elaborating why you think this? Any distribution provides the probability of the data given the covariates and parameters, at least in frequentist statistics. A larger likelihood means that the model chosen is more likely to have generated the data. This doesn't change with the distribution, and it is (or the log of it) what is used to calculate AIC. By definition, these probabilities are comparable.

And consequently, so are AIC values from different models, given that the same data are used.

@@bertvanderveen3528 stats.stackexchange.com/questions/345069/likelihood-comparable-across-different-distributions and stats.stackexchange.com/questions/139201/model-selection-can-i-compare-the-aic-from-models-of-count-data-between-linear have articulated it pretty well in these links!

@@andrebellve9027 thanks for your response! I'm afraid that a comment on stackexchange is hardly a convincing reference. My opinion on the matter remains, probabilities are comparable as they are on the same scale and measure the same thing. Regardless, it is an interesting argument, and I will look further into this. Are you aware of any scientific references on this matter? (that is, on the first link, not the second; that is hardly a good argument as the likelihood is always the probability of the data given the parameters, whatever the likelihood is). It sounds to me like that the comment merely points out that the normalising constant needs to be included, so that all distributions properly integrate to one, i.e. so that probabilities are in the range 0,1.

​@@bertvanderveen3528 I do not know of any published literature, although I can see if Kjetil would supply some references. Personally, I am not a mathematician, but I am inclined to believe him given his credentials and, as far as I can follow it, the logic is sound. Not that me following it says a huge amount! Unless I am mistaken, the original question was in regards to the use of AIC/AICc for the comparison of models with different error distributions. As AIC and it's second order derivative are defined, the values are incommensurate when the models generating them come from different distributions, providing you agree with Kjetil Halvorsen logic in the first link. It may well be that the likelihoods can be made to be comparable and some penalising condition added to make it analogous to AIC/AICc or any other information criterion, but that isn't quite the same question. Moreover, I believe (and there are a lot of packages out there so I may be wrong!) most R functions would not be altering the likelihood with a normalising constant. As such, as it stands, the AIC values outputted should not be used for comparisons when the families differ. As a side note - the second link was included for posterity; it's how I came to find the first link and it links the likelihood and AIC discussions. I know cross-validated isn't as robust as a published article, but searches in the past for formally peer-reviewed literature on the topic haven't been fruitful for me!

Hey Nic, the function matord doesn't ring any bells for me... do you know specifically what package it was from, or do you know what the function does? If the purpose is as the name suggests, to order a matrix, there is simple ways to do that in R depending on what way you're trying to order values. There seems to be a custom object within a function of the ClusterSeq package with the name "matord" but that's about all I could find rdrr.io/bioc/clusterSeq/src/R/associatePosteriors.R

Also, there's this custom function gist.github.com/pedroj/1872314

@@RJG_Ecology In order to test the relation between distance and homoplasy I create this algorithm. The general concept of algorithm is to look for the most central strain of a given group of strains. This strain is the one that minimizes the average distance within a square distance matrix. Once the most central strain has been found, the other strains are sorted in increasing distance order. Adding one strain at a time, it is possible to have an increasing number of strains coming into play. At each addition, homoplasy and average distance of the strains from the most central strains are calculated and plotted. This procedure allows to consider carefully the trend of homoplasy and distance, as well as the Rescaled Index.

@@MrAraxon Not sure if you've seen this package yet, but maybe it has some helpful functionality? www.ncbi.nlm.nih.gov/pmc/articles/PMC6412054/

Hey Margaux, yes there are two packages that are used for MLSR in R: 1) MLSTar bmcbioinformatics.biomedcentral.com/articles/10.1186/s12859-019-2887-1 github.com/iferres/MLSTar and 2) STRAIN bmcbioinformatics.biomedcentral.com/articles/10.1186/s12859-019-2887-1 I'm not very familiar with it but also 3) mlstverse github.com/ymatsumoto/mlstverse and 4) StrainR github.com/jbisanz/StrainR This blog may be helpful too: www.r-bloggers.com/2017/01/descriptive-analysis-of-mlst-data-for-mrsa/