Closed jpiaskowski closed 1 year ago
Thank you, Julia @jpiaskowski, for this proposal, this is very much appreciated. My first impression is that even if there is currently no big overlap with other task views, this might become more of an issue later on. For example, we have encouraged the proposal of a follow-up to the "Genetics" and "Phylogenetics" task views. Similarly, planning and analyzing experiments is likely to be relevant to other topics beyond agriculture. I'm not sure whether it would be possible to sharpen the focus to avoid such potential overlaps.
Thank you, I see your point.
The scope of this proposal is packages specifically developed to support agricultural research, to solve problems encountered in ag research that no other R package was/is doing. This is intended to serve a real need for ag researchers, so yes, it is broad by intention. It is not intended as a resource of packages that could be useful to agricultural research. It is our preference to not overlap with existing task views. If we do overlap (now or in the future), we can adjust as needed. But, at this time, I am trying to provide for an unmet need.
Agreed, it does feel odd to list packages like "lme4" and "nlme" that clearly have very broad usage and were not developed for agricultural research, so those could be removed. (it would be really fantastic if a CRAN task view existed for linear models, but that is outside the scope of this proposal).
The 3 packages we list for experimental design are listed in the the "ExperimentalDesign" CRAN Task View. We would prefer to repeat that info in this task view, while also indicating there is another CRAN task view covering the topic with more extensive detail. These are specialized designs invented for agricultural trials, and used largely in only agricultural trials. Listing these relevant packages enables people to find this info, but we can remove them if you think that's necessary.
The section on breeding and genetics is intended to include only genetic packages in service of plant and animal breeding, which largely concerns QTL anaylsis/GWAS, mixed models that enable heterogeneous covariance structure for genetic relatedness and genetic predictions/genomic selection models. Phylogenetics and bioinformatics are not within the intended scope. What if we called this "breeding and quantitative genetics"? That would be consistent with how the term "quantitative genetics" is used in biological literature.
Thank you for your detailed response @jpiaskowski . I would suggest that packages that are from other areas (e.g. genetics or animal science) would be included here if they are for agriculture, which should be stated in the name or description of the package, or if they contain examples of applications for agriculture (e.g. in the vignettes). Otherwise, there will be a big risk of overlap with other CTVs. Reference to agriculture in the package's name, description, or vignettes could work as a "less subjective" criterion to include the package in this CTV.
In order for that to work, there should also be a clearer definition of agriculture science besides encompasses a broad breadth of disciplines
, and one that could be understood by a non-agricultural reader.
In addition to previous comments (to which I agree), I suggest that:
Hey folks, I followed the instructions laid out in your proposal instructions but in actuality, there is a much more extensive guide that is attempting to give exactly what I thought a CRAN task view envisioned. I have been actively soliciting contributions for this from the agricultural research for the last 2 years. I did provide the link to this in my initial issue.
I do think the section "GxE" (genotype by environment) does belong in agronomic trials. There is overlap with genetic prediction, but this is separate from quantitative genetics, which is deeply concerned with prediction, kinship and specialized breeding populations. Genetics-focused journals would not publish GxE studies unless there was a substantial component accounting for genetic relatedness.
Weed science (the study of weedy plants and methods of control) and plant pathology (the study of plant diseases) are not related and should not be combined. Simply put they do different things and have different goals. The ag stats community would not want that. Certainly it is true that some sub-sections are very slim with very little to say. That is disappointing, but at the very least, this is an opportunity to communicate to those scientists what is available, even when the answer is "not much".
The definition of agricultural science is research that addresses agricultural production. The relevant topics are (also listed above):
These are subjects relevant to agriculture. They are departments located in Colleges of Agriculture at the Land Grant Institutions University of Idaho and Washington State University. Some disciplines (e.g. "horticulture") are not included for being too broad from an analytical standpoint, which is why the disclaimer "Agriculture science encompasses a broad breadth of disciplines" in included (please note that this is not a definition).
There appears to be some reticence among this group to have this CRAN task view. I can only say this is a demonstrated need among both agricultural statisticians and agricultural researchers. As an agricultural statistician (my job for the last 4 years), I have observed that many of my client do not know what is available in the R ecosystem to help them meet their research goals. And the existing CRAN task view are frequently not able to help them. For example, the former genetics CRAN task view never addressed genetic prediction and quantitative genetics. As agricultural scientists transition from SAS to R, this could be a useful resource. I am currently presenting a poster on this proposed task view at an Agricultural Statistics Conference and have received much feedback on the topic. I recently put out a blog post on this topic, which received interest on Twitter. This is a resource the agricultural research community needs.
The packages included are ones recommended by the community (e.g. "soilDB", "aqp", "apsimx"), ones in which I found reading applied plant breeding papers (e.g. "rrBLUP"), ones written by ag researchers ("StageWise", "SpATS"). There is one package not written for agricultural research, 'rnoaa', which could be removed. Otherwise, these are packages vetted for development and usage in agricultural research.
As a potential co-maintainer and having been trained and employed as a plant pathologist in the past, I second what @jpiaskowski has said about the groupings. These groupings represent a well-thought out structure that agricultural users of R will understand when looking at something like this.
Might I ask to what extent you feel that your coverage is also adequately representative of non-US agriculture? Not my field, but you only list cropScape among many other packages with names beginning with crop, and others are Australian, Brazilian, etc; the same with soils (febr). A co-maintainer for example from Brazil might be able to help broaden the proposal.
Dear @jpiaskowski : no reticence from my side (I am myself working for the French National Institute for Agricultural Research since 2013 so I know the topic well and also the needs; I am very well connected to the developers of statgenHTP for instance). So comments were just meant to provide some suggestions to make the (possible) task view as clear as possible for users. For further discussions:
thanks for the clarification of the organization of the task view and for providing a more detailed version of what you want to release;
about GxE: from a methodological (statistical) point of view, IMHO, what you have in experimental design or general analysis is intrinsically different from what you have in breedings and genetics, the latter being mainly based on linear and mixed models (and most of them linked to asreml), as GxE is (which is the reason of my suggestion);
I acknowledge the fact that my suggestions to group topics might not be relevant to what you envisioned to describe in the subsections and now that I see the TV, I can see that your sections are actually more paragraphs than sections. However, I still have the feeling that the organization (grouping) of the task view is difficult to understand (even for agricultural users) because, just reading the titles, I was not really able to understand what was the difference (for instance) between Agronomic trials and Crop modelling (it is clearer when you read content but usually the purpose of having titles is to help people navigate the topic). Of course, if I am the only one having this feeling, I'm probably wrong.
Other suggestions/questions:
On a very minor note, this is probably too early to address these issues (and your proposal is probably not in its final form) but, from a format point of view:
you have minor typos ("crop growth model(De Wit 2019)" would be better with a space before the citation), a broken link in the Link section and non consistent ways to cite articles "Diggle et al (2002)" but (De Wit 2019)" and Bruckner et al;
packages and views should be cited in the md file using the appropriate functions from the CTV package (this is probably something that you know but just in case).
Thank you for this feedback. Actions taken:
rnoaa
removed from the package list. Thank you, @rsbivand for the suggestion. It is not likely we are addressing databases/data sources outside the U.S. well. We will look into this.
Regarding additional links at the bottom of the proposal, I acknowledge this section to be far from comprehensive. Is this something that should be removed altogether until we are ready to present a more balanced and comprehensive collection of additional links? Putting together a comprehensive list will be a significant undertaking. I had previously thought this could be catch-all of extra relevant information that is not an R package, and that this did not need to be exhaustive. It sounds like I misunderstood this.
The core packages are:
nlraa
and agriCensData
)Thanks for your answers and modifications. I don't think that the Links section is meant to be exhaustive but it should not be biased toward a particular community either. If it links something, I think that it has to be something that is a clear reference for anyone in the field. Similarly, if you cite a regional resource, be sure that you can not easily find the same type of resource for other places in the world (or a worldwide similar resource). That being said, I wouldn't know if it is best to keep and enhance that section or to remove it entirely.
There might be a misunderstanding of what core packages are: they are a set of the most important packages in the field (so not all the packages related to kind of "core topic" but one package for each subsection, roughly speaking. They are referenced in the task view with the command r pkg("...", priority = "core")
(unlike the others which are referenced with r pkg("...")
).
There might be a misunderstanding of what core packages are: they are a set of the most important packages in the field (so not all the packages related to kind of "core topic" but one package for each subsection, roughly speaking. They are referenced in the task view with the command
r pkg("...", priority = "core")
(unlike the others which are referenced withr pkg("...")
).
Ah, I did not understand; thank you for the clarification. I'd like to chat more with the co-maintainers about this and solicit feedback from the community before answering this question.
Regarding the links section, it sounds like it would be best to pull that from this proposed CTV until it is a bit more balanced. I'd like to fill this out eventually, but that is a big task.
I just want to add that I like where this is going. Thank you @jpiaskowski and all, the draft of the CTV is looking great. A very minor suggestion: consider adding a link to the Contributors guide in "If you think that some package is missing from the list, please let us know."
Thank you @rociojoo for the kind comment. Yes, I will add that comment, good idea. I was travelling all last week, but now I can turn my attention to these suggestions.
Hey folks, just giving a quick update to let you know that we are still working on this. I just incorporated about 10 more packages based on community suggestions, and now we are trying to gather more feedback from the broader community on what constitutes "core packages". In meantime, "rnoaa" was removed for not being ag-specific and the links section was removed for lacking balance.
Hello. Here's what we have done since the last major communication:
[Meteor][] provides a set of functions for weather and climate data manipulation to support crop and crop disease modeling. The [agromet][] package includes a series of functions to calculate climatic and hydrological indices and statistics from tidy data. United States weather data from NOAA can be accessed with [rnoaa][]. Historic U.S. climate data from the PRISM Climate Group can be accessed with [prism][]. Data from the Copernicus data set of agrometerological indicators can be downloaded and extracted using [ag5Tools][].
If this looks like it is getting too broad, we can reassign {meteor} and {ag5Tools} elsewhere (probably databases), while {prism} and {rnoaa} would be jettisoned. However, this information is not captured in any other CTV. My goal when starting this was to help my colleagues that were having a hard time figuring out how to accomplish their research goals in R. If later a CTV is proposed that would overlap with ours, we could remove the overlapping sections and allow someone else to curate a topic. That is say, I'm not invested in this being the only CTV supporting agricultural research, I'm just trying to solve problems I know exist currently.
Question: We have received repeated suggestions to include the package {asreml} in the core packages list. This is a widely used and very helpful package for agricultural analysis; however, it neither on CRAN nor open source (you can find it here). It's currently listed in several sections in the draft CTV and there are several other packages specifically designed for manipulating {asreml} output (that are also included in this CTV). Can it be listed in the core packages list?
Please let us know if there is additional feedback. My next step is to do some serious editing for spelling, grammar, style, etc.
@jpiaskowski Thank you very much for this update.
I'm glad you got feedback from the community about the list of core packages. We did not write any rules for core packages, but I honestly think that they should be CRAN packages—but there could be exceptions. Since asreml
is not even open source I would not think it should be part of the core packages.
Points 2, 4 and 5: sounds great.
Point 3: did you reach out to the people whose names were suggested?
6: Most of these packages are not centered on Agriculture or created for agricultural use; based on their description I believe that the exception would be ag5Tools
. I'd suggest doing something similar to the Agricultural economics
section and cite the Hydrology CTV which already mentions most of these packages. What do you think?
Regarding the issue of core packages: Technically it's only possible to make CRAN packages core packages.
The reason is simply that the CRAN task views are mainly designed to list (and automatically install) packages from CRAN. Only these can be tagged with pkg()
in the .md file and only this function has the argument priority = "core"
. All other tags (github()
, bioc()
, etc.) just generate links in the HTML pages but do not extend the list of packages that can be installed automatically.
@rociojoo
Point 3: did you reach out to the people whose names were suggested?
Sorry, what names are you referring to?
Your suggestion for "agrometeorology" sounds good to me. This way we can keep the "agrometeorology" section and use that to send people to other relevant resources.
I will incorporate these changes and keep you posted when we have a spelling & grammar proofed draft.
(also, understood about asreml. It will not be part of the core packages).
@jpiaskowski I think I misunderstood While we have received community suggestions from outside those regions, we do not yet have a co-maintainer outside U.S. & Australia.
I thought you meant suggestions of maintainers, but I guess you were referring to packages.
Hi, I just wanted to write to see how things are going and if there's anything we can help with from the Editor's side.
Thank you for asking and sorry for the delay! We've been adding packages to the markdown file and doing some general clean-up. There were several relevant packages presented at the UseR! conference and I found a bunch more in the CRAN list (I now understand why you're so concerned about scope - keeping up with these changes takes a big effort).
I was going to read the documentation in more detail to get the formatting correct. Is a .ctv
file no longer required?
@jpiaskowski : Indeed, the ctv is not longer needed. Only the .md
file is needed and you can check how it is rendered using the function ctv::ctv2html
. You can also check one of the public repo at https://github.com/cran-task-views with other task view examples.
Excellent, thank you!
Hello, I am recently returned from annual leave and plan to focus on the next steps for this.
Here is a draft document that is following the recommended format for CTV's. It was also converted from a narrative style to largely bulleted points, following the style of most (perhaps all) other CTVs.
There are a few things to note:
Please let us know if you have additional advice, feedback or comments.
Thank you!
Thanks for these efforts. I just had a very brief look at the technical aspects and wanted to make a couple of quick comments:
view()
tag, e.g. r view("Econometrics")
. Then we can compute on this more easily and include an overview at the end of the task view etc.### Links
section at the very end. The "Additional Links" section is then rendered from that. (The reason is that this is not treated like a regular section but something we can compute on and use special rendering for.) For the content of that section, I would recommend to reiterate some of the important links you had included in the text.ctv::ctv2html("Agriculture.md")
and then browseURL("Agriculture.html")
See the documentation for more details.[shiny app][https://...]
should be [shiny app](https://...)
with round brackets for the URL.I will implement those changes. A check has not been run, yet, but we will.
Thanks Julia. A few minor comments first:
[Applied Statistics in Agricultural, Biological, and Environmental Sciences](Applied Statistics in Agricultural, Biological, and Environmental Sciences)
(to be replaced with Applied Statistics in Agricultural, Biological, and Environmental Sciences
or with a proper link).lme4gs
is not cited with r github...
as it should<a href="www.simplace.net">Simplace</a>
is broken (missing http
or https
I guess)r github(....)
will work. It looks like it may have been ported over from a now abandoned R-forge project. I tried to indicate this in the text, but if you have other suggestions, let me know. ctv::ctv2html()
and/or will a TOC cause problems with that command? (another question for @zeileis)Thank you for your support and guidance.
Hello Julia, Would you please remind me what I should review?
Thanks, Janet
From: Julia Piaskowski @.> Date: Thursday, August 11, 2022 at 8:37 AM To: cran-task-views/ctv @.> Cc: Williams, Janet @.) @.>, Mention @.***> Subject: Re: [cran-task-views/ctv] CRAN task view proposal: Agriculture (Issue #27)
Thank you for your support and guidance.
— Reply to this email directly, view it on GitHubhttps://urldefense.com/v3/__https:/github.com/cran-task-views/ctv/issues/27*issuecomment-1212157701__;Iw!!JYXjzlvb!h6tKjgf2-e6NF_KbJh6H70di2R6gqDf2qkntTzBbhZCqRWtpEq-Wj_wE63YiXevvSAzy4oNpO1kNmrKe2YZ4EKMAWQ$, or unsubscribehttps://urldefense.com/v3/__https:/github.com/notifications/unsubscribe-auth/AAIRUB46NKAHBNW6DU27MZDVYUM45ANCNFSM5VZXHLBA__;!!JYXjzlvb!h6tKjgf2-e6NF_KbJh6H70di2R6gqDf2qkntTzBbhZCqRWtpEq-Wj_wE63YiXevvSAzy4oNpO1kNmrKe2YYs8DlTwg$. You are receiving this because you were mentioned.Message ID: @.***>
1 Your header looks fine to me. Nathalie @tuxette I think you looked at the HTML version which was produced by ctv2html()
and hence automatically added the citation.
3 I think r github("perpdgo/lme4GS")
would be perfectly fine. That link will take viewers to the README which nicely explains what is where, I think.
7 There is no unified infrastructure for TOCs but you can look at the Official Statistics task view for an example with a short overview at the beginning. Also, the Distributions task view is an example for a more extensive TOC.
All recommended changes have been made. I also updated our checks file to evaluate package status (on CRAN or not), if the URLs work and if the date is correct.
Let us know next steps, please.
Sorry for the misleading comment on headers!! Everything looks fine from my side.
As for the next steps, if I'm right, @zeileis and myself have up-voted your proposal so @rociojoo will review your last proposal and should maybe be able to make a final decision.
I agree! Additionally, other CRAN Task View Editors are, of course, also welcome to endorse the proposal - either by commenting or by giving the previous comment from @tuxette a thumbs up. @davidjohannesmeyer @eddelbuettel @rsbivand
Posting the links here again to make it easier to find: CTV md file CTV html file
Hey folks, just checking in about this.
We already have sufficient endorsement but were waiting for the feedback from @rociojoo who, I believe, is/was traveling. Rocío, do you know when you will get round to this?
Not sure i could give my two cents, but anyway here are. I would suggest some packages for your consideration: https://cran.r-project.org/package=FielDHub / https://github.com/DidierMurilloF/FielDHub A shiny design of experiments (DOE) app that aids in the creation of traditional, un-replicated, augmented and partially-replicated designs applied to agriculture, plant breeding, forestry, animal and biological sciences. *Category: Experimental design
**Not on CRAN / https://github.com/OpenDroneMap/FIELDimageR FIELDimageR: A Tool to Analyze Images From Agricultural Field Trials and Lab in R. Manipulation of Multispectral images, plant part count. Category: High throughput phenotyping (HTP)
**https://cran.r-project.org/package=photosynthesis / https://github.com/cdmuir/photosynthesis package with modeling tools for C3 photosynthesis, as well as analytical tools for curve-fitting plant ecophysiology responses. Category: Crop growth models & crop modelling
**https://cran.r-project.org/package=tealeaves / https://github.com/cdmuir/tealeaves package to model leaf temperature using leaf energy balance, companion of photosynthesis Category: Crop growth models & crop modelling
**https://cran.r-project.org/package=plantecophys / https://github.com/remkoduursma/plantecophys package that bundles a number of tools to analyze and model leaf gas exchange data. Category: Crop growth models & crop modelling
**Not on CRAN / https://github.com/jstinzi/plantecowrap add to capabilities to 'plantecophys' include temperature responses of mesophyll conductance (gm, gmeso), apparent Michaelis-Menten constant for rubisco carboxylation in air (Km, Kcair),and photorespiratory CO2 compensation point (GammaStar) for fitting A-Ci or A-Cc curves for C3 plants, ability to fit the Arrhenius and modified Arrhenius temperature response functions for maximum rubisco carboxylation rates (Vcmax) and maximum electron transport rates (Jmax) Category: Crop growth models & crop modelling
**https://cran.r-project.org/package=bigleaf / https://bitbucket.org/juergenknauer/bigleaf/src/master/ package for the calculation of physical (e.g. aerodynamic conductance, surface temperature) and physiological (e.g. canopy conductance, water-use efficiency) ecosystem properties from eddy covariance data and accompanying meteorological measurements Category: Crop growth models & crop modelling
Not on CRAN / https://github.com/rasenior/ThermStats* and *https://cran.us.r-project.org/package=Thermimage / https://github.com/gtatters/Thermimage First is R package addresses current constraints on applying thermography in ecology, by speeding up and simplifying the extraction of data from (FLIR) thermal images, and by facilitating the calculation of different metrics of thermal heterogeneity for any gridded temperature data. Second is a collection of functions for assisting in converting extracted raw data from infrared thermal images and converting them to estimated temperatures using standard equations in thermography. Provides an open source proxy tool for assisting with infrared thermographic analysis. Both require the external software ExifTool or, easier for R newbie IMHO, https://github.com/JoshOBrien/exiftoolr Category: Crop growth models & crop modelling
*Not on CRAN / https://github.com/poppinace/tasselnetv2plus A Fast Implementation for High-Throughput Plant Counting from High-Resolution RGB Imagery. Some functionalities could be similar to FILEDimageR, but with completely different approach. Category: High throughput phenotyping (HTP) https://cran.r-project.org/package=DataExplorer / https://github.com/boxuancui/DataExplorer Exploratory Data Analysis (EDA) Category: Trial analysis
https://cran.r-project.org/package=ggstatsplot / https://github.com/IndrajeetPatil/ggstatsplot Not on CRAN / https://easystats.r-universe.dev/ui#packages https://cran.r-project.org/package=rstatix / https://github.com/kassambara/rstatix all 3 above pkgs for simple and intuitive pipe-friendly framework, coherent with the ‘tidyverse’ design philosophy Category: Trial analysis
Hope it helps Massimiliano
On 29/08/2022 23:11, Achim Zeileis wrote:
We already have sufficient endorsement but were waiting for the feedback from @rociojoo https://github.com/rociojoo who, I believe, is/was traveling. Rocío, do you know when you will get round to this?
— Reply to this email directly, view it on GitHub https://github.com/cran-task-views/ctv/issues/27#issuecomment-1230865595, or unsubscribe https://github.com/notifications/unsubscribe-auth/AOYDPH4Q4HEASZJ4ESBXK7DV3URPHANCNFSM5VZXHLBA. You are receiving this because you are subscribed to this thread.Message ID: @.***>
-- This email has been checked for viruses by Avast antivirus software. www.avast.com
Thanks @ScymnusRIP ! I filed an issue for these.
OK, I think we have waited long enough. We can still incorporate further feedback and suggestions later on, after the first version was published on CRAN.
Julia @jpiaskowski , could you please do the following:
ctv-agriculture
repository: Agriculture.html
(will be hosted on CRAN only) and .github
(will use CoC from CTV initiative) can be deleted. The additional_links.md
is probably also obsolete now.Afterwards I will:
cran-task-views
organization.Thanks!
Hi all, Apologies for the silence. Traveling + tropical storm + hurricane + preparing new trip starting this weekend. It's all good on my side. Just a typo (should be Bioconductor instead of Bionconductor in #Breeding).
@zeileis I will do what you requested this week.
Wonderful, looks good. I did a few minor touch-ups and standardizations: https://github.com/cran-task-views/Agriculture
I think we're ready to go. If you approve, I'll release the task view on CRAN.
Scope
Agriculture science encompasses a broad breadth of disciplines. Many many package in base R and contributed packages are relevant to agricultural researchers. For that reason, this is not exhaustive list of all packages useful to agriculture researchers. It is intended to cover major packages that in most cases, have been developed specifically to support agricultural research and analytical needs. We intend this as a resource for agricultural scientists and agricultural statisticians.
Our core areas encompass packages for:
Packages
We have already drafted an agricultural task view here in a git repo (although no CTV file has been prepared). The majority of these packages are on CRAN, but a few can only be found on GitHub and other alternative repositories.
Here is an alphabetised list of packages:
AGHmatrix
agricolae agridat
agriTutorial
agroBioData
apsimx aqp asremlPlus
bayesammi
BGLR
breedR
cdlTools
CropScapeR
desplot
diaQTL
drc
DSSAT
emmeans
EnvRtype
epifitter
epiphy
fabio
FedData
fertplan
ggfertilizer
gge
GWASpoly
hagis
hnp
IBCF.MTME
INLA
LinkageMapView
lme4
lme4gs
lme4qtl
mappoly
MapRtools
MCMCglmm
MegaLMM
meteor
nlme
nlraa
pedigreemm
phenorice
phenoriceR
polyBreedR
polymapR
polyqtlR
poppr
PROSPER
qgtools
qtl
qtlpoly
rapsimng
rarms
Recocrop
rMVP
rnassqs
rnoaa
Rquefts
rrBLUP
rusda
Rwofost
sharpshootR
simplePHENOTYPES
sommer
soilDB
SoilTaxonomy
SoyNAM
StageWise
statgenGWAS
statgenGxE
statgenHTP
statgenIBD
statgenMPP
statgenSTA
tidyUSDA
usdampr
Overlap
In several cases, this proposed task view references other existing task views (e.g. spatial, econometrics) when those task views were the best alternative to repeating information. There is a small amount of overlap between the agricultural databases packages in this proposed CRAN task view and Official Statistics (e.g. "cdlTools", "FAOSTAT").
In general, there does not appear to be substantial overlap with existing CRAN task views.
Maintainers
principal maintainer: Julia Piaskowski (@jpiaskowski)
possible co-maintainers: Janet Williams (@janetw), Adam Sparks (@adamhsparks), Andrew Kniss