UPDATE: pkg API has changed - updated the post below to work with the current CRAN version, submitted 2016-08-02 I was at a hackathon focused on Ocean Biogeographic Information System (OBIS) data back in November last year in Belgium. One project idea was to make it easier to get at data based on one or more marine regions. I was told that Marineregions.org is often used for shape files to get different regions to then do other work with. ...

We (rOpenSci) just held our 3rd annual rOpenSci unconference (https://unconf16.ropensci.org/) in San Francisco. There were a lot of ideas, and lots of awesome projects from awesome people came out of the 2 day event. One weird idea I had comes from looking at the Node world, where there are lots of tiny packages, instead of the often larger packages we have in the R world. One reason for tiny in Node is that of course you want a library to be tiny if running in the browser for faster load times (esp. on mobile). ...

GenBank IDs, accession numbers and GI identifiers, are the two types of identifiers for entries in GenBank. (see this page for why there are two types of identifiers). Actually, recent news from NCBI is that GI identifiers will be phased out by September this year, which affects what I’ll talk about below. There are a lot of sequences in GenBank. Sometimes you have identifiers and you want to check if they exist in GenBank, or want to get one type from another (accession from GI, or vice versa; although GI phase out will make this use case no longer needed), or just get a bunch of identifiers for software testing purposes perhaps. ...

GitHub issues are great for humans to correspond over software, or any other project. At rOpenSci we use an issue based software review system (ropensci/onboarding). Software authors and reviewers go back and forth on the software, making a better product in the end. We have a relatively small number of pieces of software under review at any one time compared to e.g., scientific journals - however, even with the small number, we as organizers, and authors and reviewers can forget things. For example: ...

scrubr is an R library for cleaning species occurrence records. It’s general purpose, and has the following approach: We think using a piping workflow (%>%) makes code easier to build up, and easier to understand. However, you don’t have to use pipes in this package. All inputs and outputs are data.frame’s - which makes the above point easier Records trimmed off due to various filters are retained as attributes, so can still be accessed for later inspection, but don’t get in the way of the data.frame that gets modified for downstream use User interface vs. speed: This is the kind of package that surely can get faster. However, we’re focusing on the UI first, then make speed improvements down the road. Since occurrence record datasets should all have columns with lat/long information, we automatically look for those columns for you. If identified, we use them, but you can supply lat/long column names manually as well. We have many packages that fetch species occurrence records from GBIF, iNaturalist, VertNet, iDigBio, Ecoengine, and more. scrubr fills a crucial missing niche as likely all uses of occurrence data requires cleaning of some kind. When using GBIF data via rgbif, that package has some utilities for cleaning data based on the issues returned with GBIF data - scrubr is a companion to do the rest of the cleaning. ...

request is DSL for http requests for R, and is inspired by the CLI tool httpie. It’s built on httr. The following were driving principles for this package: The web is increasingly a JSON world, so we assume applications/json by default, but give back other types if not The workflow follows logically, or at least should, from, hey, I got this url, to i need to add some options, to execute request - and functions support piping so that you can execute functions in this order Whenever possible, we transform output to data.frame’s - facilitating downstream manipulation via dplyr, etc. We do GET requests by default. Specify a different type if you don’t want GET. Given GET by default, this client is optimized for consumption of data, rather than creating new things on servers You can use non-standard evaluation to easily pass in query parameters without worrying about &’s, URL escaping, etc. (see api_query()) Same for body params (see api_body()) The following is a brief demo of some of the package functionality: ...

The first version of binomen is now up on CRAN. It provides various taxonomic classes for defining a single taxon, multiple taxa, and a taxonomic data.frame. It is designed as a companion to taxize, where you can get taxonomic data on taxonomic names from the web. The classes (S3): taxon taxonref taxonrefs binomial grouping (i.e., classification - used different term to avoid conflict with classification in taxize) For example, the binomial class is defined by a genus, epithet, authority, and optional full species name and canonical version. ...

I gave two talks recently at the annual Crossref meeting, one of which was a somewhat technical overview of programmatic clients for Crossref APIs. Check out the talk here. I talked about the motivation for working with Crossref data by writing code/etc. rather than going the GUI route, then went over the various clients, with brief examples. We (rOpenSci) have been working on the R client rcrossref for a while now, but I’m also working on the Python and Ruby clients for Crossref. In addition, the Ruby client has a CLI client inside. The Javascript client is worked on independently by ScienceAI. ...

I maintain an R client for the GBIF API, at rgbif. Been working on it for a few years, and recently been thinking that there should be a nice low level client for Python as well. I didn’t see one searching Github, etc. so I started working on one recently: pygbif It’s up on pypi. There’s not much in pygbif yet - I wanted to get something up to start getting some users to more quickly make the library useful to people. ...

I’ve recently made some improvements to the functions that work with ISD (Integrated Surface Database) data. isd data The isd() function now caches more intelligently. We now cache using .rds files via saveRDS/readRDS, whereas we used to use .csv files, which take up much more disk space, and we have to worry about not changing data formats on reading data back into an R session. This has the downside that you can’t just go directly to open up a cached file in your favorite spreadsheet viewer, but you can do that manually after reading in to R. In addition, isd() now has a function cleanup, if TRUE after downloading the data file from NOAA’s ftp server and processing, we delete the file. That’s fine since we have the cached processed file. But you can choose not to cleanup the original data files. Data processing in isd() is improved as well. We convert key variables to appropriate classes to be more useful. isd stations ...