At the recent Discovery On Target conference, a workshop on data and analytics for drug discovery contained several informative talks. This week’s Video Tip of the Week was inspired by the first speaker in that session, Georges Grinstein. Not only was the software he talked about something I wanted to examine right away (Weave)–his philosophy on visualization of data was so in line with my informal thoughts on the topic that I just connected with it immediately. But also–stay for the “living figures” down below.
Grinstein has been working on dataviz for a long time. And he’s been working with big data since long before big data was trendy. For some of his background and philosophy, check out this talk at a VIZBI conference. Because so many of the problems are the same across big data types, the software that he’s been working on could really be useful for the new issues facing big data in biology. But I don’t know that I’ve heard about it among the genoscenti just yet. (In this talk he also covers RadViz, a radial visualization tool that some folks might find useful. It was also mentioned in the workshop.)
One of the key things that he wanted us to take away from the workshop was that we need to offer people multiple, interactive visualizations for them to get the most of out the data. This is something I’ve been looking for quite a bit. I fell in love with an early version of the Caleydo stuff for exactly this reason. But I understand that it can be tricky.
Weave, or the Web-based Analysis and Visualization Environment, gets closer to this with super responsiveness than I’ve seen elsewhere. This week’s Video Tip is a short intro to this platform, but I’ll link you below to a longer form that you should watch if you want to dive into this tool. Here you’ll see that just by dragging a CSV file in, you can then set up a scatter plot, bar chart, parallel coordinates, a color histogram, and a table. In seconds. Really.
This brief intro doesn’t do full justice to this tool, of course. I joined the Weave-users discussion group and found a recent webinar recording that you should watch. But you’ll have to grab it from the group, it doesn’t appear to be stored on a video platform site (search for the thread called “IVPR Update on Weave Monday 3/23“). It goes into more detail on the features, of course. And sharing data, and reproducibility of the visualizations with the session history options.
I downloaded the Weave Desktop and ran it on my little system. I grabbed some transcription factor score data from the ENCODE project with the UCSC Table Browser, got it in csv format, pulled it in, and within seconds was looking over all the data on the X chromosome for this TFBS I was interested in. Clicking an item in my table highlighted it in my histogram. And that was just to kick the tires. According to the video, you could have had a tile of Cytoscape (because you can integrate with Cytoscape–I didn’t get that far yet though) and checked out interaction data as well. Although I mention Cytoscape because readers here probably know it, that’s just one of the linkable tools. R is embedded, and other stats tools, and you can modify your scripts right from Weave. Some of these additional features may be part of the Analyst Workstation sub-project. I couldn’t always tell which tool had which features in my early explorations.
But if there’s one thing I’d like you to do after reading this post (if you read this far) is look at this paper that is just out. As I was noodling on Weave, I thought to myself that it was PERFECT to create the kind of “living figures” that I want to see in more papers. Now go see Dynamic Data Visualization with Weave and Brain Choropleths. I don’t care if you aren’t interested in brain choropleths–go look at the figures. In each one, there’s a link to a Weave demo, like this:
Click on those demos to load them. You can be interacting with the data on the brain maps, with pre-set Weave tiles of different features of the data set for you. Open the gears icons to change the settings. Now imagine this with gene expression maps in C. elegans bodies. Or with transcription factors and scores in mouse embryos. Or Venns with big piles of GO terms (but what I really want there is UpSet anyway). Or any of a dozen other types of data we get in big data papers now that are really impossible to explore in traditional publication format. I want this for genomics papers in the future, okay?
This software has a lot of potential for analysis, visualization, and sharing of data. I can’t cover it all in a brief blog post. The Weave team has thought carefully about sharing with colleagues, reusable templates, and provenance of data, and all this is built right into to this tool. If you are analyzing data for others, you can set up dashboards for them to see specific views. See their help and info docs for more details, and check out the longer videos in the forum. I think it would connect with a lot of people–and could benefit the genomics community greatly. Have a look. I think you’ll like it.
Weave-users discussion: https://groups.google.com/forum/#!forum/weave-users
Weave desktop: http://info.oicweave.org/projects/weave/wiki/Installer
More videos, Weave IVPR channel: https://www.youtube.com/channel/UCXJrO9cug3c7B7eRJSwZ4vQ
Patterson, D., Hicks, T., Dufilie, A., Grinstein, G., & Plante, E. (2015). Dynamic Data Visualization with Weave and Brain Choropleths PLOS ONE, 10 (9) DOI: 10.1371/journal.pone.0139453
Daniels, K., Grinstein, G., Russell, A., & Glidden, M. (2012). Properties of normalized radial visualizations Information Visualization, 11 (4), 273-300 DOI: 10.1177/1473871612439357