Penflip, the writer-friendly GitHub

Screen Shot 2014-06-28 at 4.04.52 PMThe average number of authors per research papers has been steadily increasing over the past few decades. Naturally, writing papers became more collaborative, each researcher contributing to their area of expertise. But collaborating on a document is not always easy. Multiple Word documents sent back and forth by email,  text highlighted in all the colors of the rainbow to track who changed what in the text… Lucky for us researchers, over the past few years, a number of online tools have been developed to make collaborative writing a pain-free experience.

Inspired by the GitHub platform that allows programmers from around the world to collaborate on software code, Loren Burton developed Penflip. Penflip provides a web-based text editor that includes basic formatting options (bold, italic, lists, links..). You will not find advanced formatting options commonly found in other editing software such as Word or Google docs. But Penpile’s strong point lies somewhere else:  its advanced version control system.

In GitHub, large chucks of code can be extracted from a program, edited, then accepted back into the program while keeping track, line per line, of changes made. Github could in principle be used for text editing, but the whole platform has been designed for coders, with technical jargon that will scare many away. A writer-friendly version of GitHub was needed: Penflip.

Here is what the interface looks like. Pure and simple.

Screen Shot 2014-06-28 at 4.14.21 PM

After each modifications, additions are marked in green, and subtracted text is crossed out.

Screen Shot 2014-06-28 at 4.15.48 PM

Penflip offers many interesting features such as:

  • The ability to integrate mathematical formulas using MathJax.
  • Chapter management that allows to create chapters that can be easily drag & dropped to reorder.
  • Inline comments to discuss specific pieces of the text.
  • Discussion tools to share ideas and comments.

The free version of the tools is limited to public documents, meaning anyone could access your articles. A $8/month fee will give you access to private documents and premium support.

Note: another GitHub for writer is SciGit. The service is on hold for the moment, but keep an eye out for its return.

Extracting data from plots, images, and maps with WebPlotDigitizer

Screen Shot 2014-06-12 at 1.14.41 PMIt is hard to understand why in today’s all-interactive world, scientific data continues to be represented as still, lifeless images. Sadly, the good old pdf or html versions of research papers do not allow us to extract and reuse the data represented as graphics. While we wait for publishers to finally bring interactive figures to publications, and for data to be more open an easily accessible, a simple and reliable method to extract data from graph would be great. Lucky for us, WebPlotDigitizer does exactly that!

WebPlotDigitizer is a free, online tool, that allows anyone to upload the image of a graph and extract the data. Here are the announced features of the version 3.2 (released in May 2014):

  • Web based. No installation needed.
  • Supports XY charts (even skewed and non-orthogonal), polar plots, ternary diagrams and maps.
  • Automatic curve extraction algorithms aid rapid extraction of a large number of points.
  • Generates data in .CSV format which can be used by any data analysis program like Excel, OpenOffice, Origin etc.
  • A zoomed in view on the side aids accurate selection of data points.
  • Free of charge and distributed under the GNU General Public License Version 3.

This is how it works. First load your the image image of the graph by a simple drag and drop from the webpage or by browsing through your files on your computer. The image can be a screenshot, or even a photograph captured by your webcam since WebPlotDigitizer supports .jpeg, .png, .bmp and .gif. You can then make simple adjustments to the images, flipping or cropping it to your liking. The first step of the analysis consist in defining the type of graph analyzed and calibrating the axis by assigning four points of known values on the axis.

Then comes the extraction of the data points. A manual and an automatic mode are available. In the manual mode, you click directly on the graph to add data points and WebPlotDigitizer calculates the precise coordinates of each point. Simple and reliable but can be time consuming.

The automatic mode uses an extraction algorithm that will automatically identify and extract data points from the image. A few of parameters must be played around with to pinpoint the line or points that you wish to extract. The data can then be exported as a .cvs file or directly plotted using Plot.ly, a web-base graphing tool. The integration of WebPlotDigitizer in Plot.ly really differentiates this tool from the many other solutions, making plotting and sharing the extracted data incredibly easy.

Here are a few useful links to get you started:

And do not hesitate to make a donation to Ankit Rohatgi, creator of this wonderful tool, go to his website and click on donate!

Roll the dice! PIpredictor calculates your likelihood of becoming a PI

qubodup_Ugly_non-perspective_cartoony_fort_fortress_stronghold_or_castle-1Academia is like a middle age fortress under siege, set in stone by traditions and habits, and ever so hard to get into. Whether academic life is a good choice for you or not is one question. But even if you want to be a Principal Investigator (PI), only between 0.45% and ~20% of PhD students end up in a tenure track position or similar. So you might think your odds are not great.

Well, now there is a tool to actually calculate those odds. Lucas Carey (Pompeu Fabra University) and his coworkers used machine learning to extract a set of laws that determines your chances of becoming a PI. This was achieved by looking at the careers and publication records of over 25,000 scientists in the PubMed biomedical literature database. They found that access to PI positions is highly predictable based on publication records and other parameters such as gender.

The study was recently published in Current Biology and a web-based tool, named PIPredictor, allows anyone to calculate their likelihood of becoming a PI. Any PI wannabe, should indicate their name, gender and a list of publication. The tool is free, but requires your publications to be referenced in PubMed.

PiPredictor allows anyone to calculate the likelihood of becoming a PI based on his/her publication record.

PiPredictor allows anyone to calculate the likelihood of becoming a PI based on his/her publication record.

A fun tool indeed! But beyond the amusing aspect of this study, the predictability achieved here is a reminder of the rigidity of academic institutions. In particular, the conservative approach to recruiting by many universities and research agencies. Lets (naïvely) hope this will be changes with the newer generations of academics.