Articles by Bill Flanagan

Bill Flanagan’s background is in building online collaboration systems.

It’s been a while since my last update. But lest anyone have any doubts about the ongoing direction of OpenWetWare, rest assured, things are moving forward.

Not dead yet….

Server Move Details

This week (June 1 – June 5), will be moving from our current server at Rackspace to a new Rackspace Cloud Server. The server will be around the same class of machine and will be running on Ubuntu Linux rather than the existing RedHat Enterprise Linux release. All backups will be done to Rackspace Cloud Files. All MySQL database backups , and image files will be stored external to the server via Cloud Files. For those of you who are wondering why OWW will be using Ubuntu rather than Red Hat, it’s because Wikimedia uses Ubuntu for all of their MediaWiki servers; using it will keep OWW close to the infrastructure that MediaWiki is tested and developed  on.

The move will be done Tuesday night around 11:00 PM EST. We don’t anticipate problems but the server will briefly go down as the IP address is changed. The new server has been configured and, just after changing the IP address, the most recent snapshot of the MySQL databases from the current server will be loaded to the new one and a final file sync will be executed.

There should be no changes in the way MediaWiki and any extensions are handled. LaTex has been installed on the new server.  All extensions are working or are being tweaked before the move.

There will be no upgrade of OWW’s MediaWiki software release until the move is complete. Hard-won experience dictates that reducing variables is the right way to maximize the probability of a successful major server task.

Since OWW uses many virtual hosts, all of these will be tested briefly to make sure they are all accessible. This can’t be tested completely until the change. No problems are anticipated but if there are problems, this is the most likely place it will be.

Please submit any comments or questions to me. Either reply here or use this link and follow the contact instructions in the OpenWetWare wiki.

Thanks again.

Bill Flanagan

To lookup an article using a document object identifier, there’s a cheap and cheerful way to do it based upon the work we did earlier to add access to pubget.

Without it, you can always resolve a DOI. Here’s a simplest example: Let’s say the DOI is 10.1021/ac7018574. This is an article by Cameron Neyland. If you want to redirect t the document, you can use an open DOI like this:

You can therefore represent a DOI  in OpenWetWare like this:

[ doi:10.1021/ac7018574]

That would display the string, doi:10.1021/ac7018574. Clicking on the link would bring you to the paper. This won’t provide any information about the document. Clicking on it will redirect your browser to the paper.

In addition to PubMed lookup, Pubget also includes a DOI resolver. You can access it via an RSS feed within OpenWetWare.

I’ve added a page including a set of examples and instructions to OpenWetWare to illustrate how this can be done as well as how the previous examples. You can look at the list of examples (and, please, add more!) here:

The result is that the title of the doc and a link to allow you to read the document are displayed. There’s currently no way to format these string directly but it’s coming. Look at the page to see an example. I’ll simplify this to allow an even shorter syntax to display it.

Just as pubget uses information about a university to determine what periodicals you have access to, the DOI system uses a ‘resolver’ to map a DOI into a reference to the periodical and the document described.

Biblio, the reference/citation extension in OpenWetWare currently does not support DOI’s. There is limited support for importing the metadata such as periodical, edition, authors, and even title, to do this withougt a bit of hacking. Let me know if anyone is interested in seeing this support added.

If anyone has questions about the use of a feature like this, please let me know. As I mentioned, don’t hold back on adding examples to the page in OWW.


MediaWiki is the software that is built on. We customize it by applying out own styling to the page, add out own member management software to it, and either write our own extensions to it or download and install others. In general, we try like hell to not touch the ‘core’ MediaWiki code. It would be like a lab scientist starting with a standard protocol and then making many modifications to it for his or her own purpose. That’s OK until someone makes a set of modifications to the protocol you started with. You now have to figure out how to retrofit your procedures, ingredients, and what-not with the new ones. Sometimes it’s easy. But sometimes it’s just too much and you revert back to the original, throwing out your own potentially useful enhancements.

MediaWiki has a number of built-in features. One of them is the “magic link” feature I mentioned in the last message. For some reason, someone added a feature to MediaWiki that, when a specific keyword was found, it would try to see if the next word was recognizable and do something on that basis. In the case I discussed, it was the “PMID” word followed by a recognizable PubMed number. The rules for such a number are very simple to understand. But they are not always easy to make work reliably. Since any series of numbers will be seen as a PMID, the wiki page will now contain a link to the PubMed database, offering users the chance to see the citation for a paper. If the paper doesn’t exist, the lookup will still take place. As the saying goes, “Garbage in, garbage out”. The change we made was a bit of a “core” change, but we’ve already moved it into an ‘extension’, thus allowing us to move to a new MediaWiki version without losing our ‘core-like’ change.

We removed another built-in document identifier that was built into MediaWiki. This one was ‘RFC’. The term came from the Internet. All aspects of Internet standards are documented in a “Request For Comment”. Biopart RFC standards are based upon similar documents and processes. In this case, we removed the RFC “magic links” to keep OpenWetWare for turning references to Biopart RFC’s into Internet RFC’s.

MediaWiki also supports magic links for ISBN number lookup as well. Again, the very simple syntax of the number itself makes this possible.

One identifier that’s conspicuously absent from this list of global standards is the academic publishing world’s standard, the DOI.The Document Object Indetifier. The standard has a is different from some of the other schemes. For one thing, you need to pay to obtain a license to issue them. Another is that you also, as far as we have ever determined, need to pay for each DOI issued. Unlike the simple, regular nature of the PMID, the DOI label itself is a variable string that isn’t easy to handle. Once you have located a DOI in a document, to resolve the publication it came from, a third-party needs to step in to do the actual lookup.

In the US, the PubMed database is administered by the NIH. Any PMID can be looked up via a single interface. The simplicity of this is what allowed Pubget to step in and cut out the NIH by directing a reader with a valid subscription to the actual text of a PMID-identified reference.

You can do something of the same thing with the a DOI, if the content of a string can be identified clearly as a DOI. At some point OWW may add support for the DOI in this manner. That would mean that Biblio references could contain them as well as any OWW string that either started with “doi:” or followed the word, “DOI” would lead to the article itself. Like Pubget, the resolution of a DOI to a particular publisher and their online repository for articles does require a subscription to the periodical. Another use of the DOI would be OWW issuing a DOI for a page. This bears more explanation. But also more research.

I’ll provide a simple example of a DOI lookup via a resolver if people are curious.



To paraphrase Sigmund Freud,

“Sometimes a reference is just a reference…”

But not all the time. By the time we’re reading an OpenWetWare (OWW) page and notice a reference to a published paper, we’ve come to the page through a search engine. At this point, a reference is no longer a starting point for a new search. It’s now contextually enlivened information.

We want to read it now. Our interest has already been qualified. Searching is behind us. Is this the time to go to another site to read about the article and to find more? For me, at this point, I want to decide how relevant an article is by reading it. “Let me decide!”.

PubMed’s website is where OWW references typically end up. The PubMed ID (PMID) is a common identifier for research articles for most biological research output. The US NIH acts as a proxy for redirecting us to the actual article. If we have rights to read it via online subscription, we eventually get to see it.

As we’ve experienced far too often, getting to the article through PubMed isn’t direct. The process looks something like this:

  1. You see a citation.
  2. You want to read it.
  3. You click the link.
  4. You read the reference.
  5. You click the link.
  6. You go to the publisher’s site.
  7. You click the link.
  8. If your institution has a subscription, you read the article.

Enter Pubget.

Pubget is a service that cuts through this process. It allows you to click on a link and read the article.

Using it, the process gets collapsed to this:

  1. You see a citation.
  2. You want to read it.
  3. You click the link.
  4. If your institution has a subscription, you read the article.

Note the steps. Once click is now required. The rest is all about thinking, wanting, and reading. So. What if some of the power of this service were to be brought to bear within OpenWetWare?

Based upon a request last week, we looked into how Pubget works. With no official affiliation with Pubget, I’ve added a number of features to OWW that now make this possible in an almost invisible manner.

What do you have to do to take advantage of Pubget

If you’re a member of OWW, you can start by going to your preferences page to set your institution. Just log in, click on the “My Preferences” link at the top of the page, and go to the ‘Misc’ tab. It’s the last tab on the right. Look for the list of participating universities here:

Select the name of the institution where your lab is located. Like Pubget itself, this feature will only work when you’re physically working on that network. Once you set this, you need never change it until you graduate! If you’re at home and not at work, pubmed will use the default setting: any subscriptions by your university will be unavailable util you get back to the lab. This isn’t part of OWW’s Orwellian plot to keep you at your bench for as much of your life as possible. It’s a technical issue. At MIT, we use browser certificates that extend this so that I still get access to the same content at home. I’m sure this isn’t universal. But try it and see how it works. We’re evaluating how we can extend this capability over time for other universities. If your university uses VPN connections, you are virtually on your lab’s network: all subscriptions will continue to work.

Even if your university isn’t on the list, this feature will allow you access to the full text of all open source journals such as those published via PLoS.

The effect is that a growing number of OWW lookup features will use this label to pre-set the URL’s you use to ‘cut to the chase’ and make content available when you click and not a few seconds later. If you want to review several articles, those few seconds do add up quickly. In addition to this, Pubget provides a search interface that ‘knows’ the format of many of the most important journals.

What parts of OWW can use Pubget?

General OWW text.

Try this. Edit an OWW article you’ve created containing a pubmed reference. Reformat the reference to use this format:

PMID 12345

Save the document. The text is now a clickable reference. If you’ve set your preference, the article will be opened for you to read if it’s available as a PDF file as well as associated reference information. You can use the PubMed search to find related content as well. Any existing pages using the ‘PMID’ tag will beave the same way.

Biblio References

If you use the Bibio extension for publishing citations, a Pubget link is now present if there’s already a pubmed reference for it. Just click on the link to to read the text.

Pubget Gadgets

A Pubget gadget is a code snippet you can create at the pubget site to provide a search using their service. To create one, go to the pubget site and create it using this url:

To paste it onto your page, do the following:

  1. Create the widget on the Pubget page.
  2. Copy the widget to your clipboard after you create it.
  3. Paste the html code onto an OWW page.
  4. Place an  “<html>” tag right before the code you pasted.
  5. Place an “</html>” tag right after the code you pasted.
  6. Save the page.
  7. Instant widget!

RSS Feeds

A side effect of the way pubget works is that RSS feeds, already supported within OWW, can use the same feature. To create and RSS feed that uses Pubmed, do the following.

  1. Open one of your OWW pages for editing.
  2. Create an RSS feed using the with a search term inside of it like this:<xfeeds>OpenWetWare</xfeeds> (replace OpenWetWare with your text)
  3. Add the ‘bubget’ attribute (pubget=”Y”) inside of the “<xfeeds> tag. The sequence will now look like this:   <xfeeds pubget=”Y”>OpenWetWare</xfeeds>
  4. Save the OWW page

This will now tell the xfeeds extension in OWW to map all of the URL’s in the returned RSS feed to use your specified university tag. The references will now direcly load when you click on them.

There’s a shortcut to do this if you have a page name you want to use as the search term. As an example, consider your user page.

  1. Edit your user page.
  2. Find a place you want to insert the search feed.
  3. Insert the following tag: “{{PubgetRSS}}” (no quotes but with the curly braces)
  4. Save the page. Done!

If anyone sees other places this capability would fit, please let me know.

Also, if anyone has a problem using this feature, let me know. I’m not trying to push one company versus another. But the service to OWW this offers seems worth the effort.



My life at OWW has been an endless stream of messages articulating Austin’s far-too-old feature and technology suggestions that I slowly get around to adding. The “flash” (of insight) to “bang” (of getting the idea online) is not great; I would hope the time will diminish eventually. But for now, this is what it is!

The latest is a big one. That would make it a “big bang”. So maybe a better way of saying it would be a “marginal thud” to a “moderate drop”. I’ve started the process of adding support to OWW for Javascript Gadgets. This is a centrally managed method of deploying ‘sanctioned’ Javascript that then can be enabled or disabled by every user. This extension is already in use on various WikiMedia servers; we are long overdue.

What it means is that there are scores of ways people more javascript-savvy than me have created small extensions to MediaWiki that do all sorts of useful things. The ‘quick-nav’ item on the sidebar, written by Austin, as an example, could be included in this general category. I created an extension to enable it for everyone since it is so useful to anyone who has ever forgotten the last 15 pages he/she visited.

I’ll provide a full list of these extensions when I’ve completed the import of them.

Once we have access to all of them, I would imagine that a set of much more research- and life science-specific entries would be useful. Since the Venn diagram representing “biologists”, “OWW users and viewers”, and “Javascript hackers” may be initially limited to Austin, it may be a while before we have many OWW-specific extensions available.

Personally, I think we should all do our best to start keeping him company.

If there’s something you have to do over- and over- again in OWW to do your work, consider using the discussion area I’ll add in order to get the ideas flowing.

Expect the first set of Gadgets, with instructions, to be available this week.

If anyone wants to volunteer to help out with testing Gadgets prior to our including them in the central library, please let me know. We’re not limiting inclusion of Gadgets because we want to suppress open science, by the way. It’s just that in programming, anything that can fail, will. I just don’t want an infinite number of new lab notebook pages to be created just because someone wanted to automate his or her own task and didn’t test!

Here’s a link for more information on MediaWiki Gadgets: Gadgets

By the way. Don’t confuse Gadgets with Widgets. We may add Widgets as well. Unlike Gadgets, once enabled, Widgets can be added by anyone to any of their pages. Where Gadgets are more related to creating content and using OWW, Widgets will be useful for extending OWW to interact with external data.


PS: Thanks again, AC.

And… graduates with his PhD in Biological Engineering from MIT. Jason was joined by other OWW stallwarts including Dr. Reshma Shetty, Dr. Barry Canton, Dr. Ty Thomson, and Dr. Samantha Sutton as they completed their degrees.

Dr. Jason Kelly

Congratulations to all of them.

NPR’s RadioWorkshop ran a segment on Tuesday’s Morning Edition on the MIT 2006 iGEM team’s work on a banana and mint scented E. coli (aka Eau d’Coli) … all done on OWW. The report featured a graduating senior, Stephen Payne, and OpenWetWare’s own (also soon to be graduating with her doctorate degree!) Reshma Shetty.

From the NPR page:
MIT Eau D'Coli Team

Even if you’re already familiar with the results, E. coli which has shed its rather distinctive scent in exchange for mint and banana, the RadioWorks report is well worth a listen.

In addition to the great report, a Biological Engineering “theme song” is sung at the end of the report in the style of a Gilbert and Sullivan ditty.

Here’s a link to the story and the downloadable audio:

From the report:

“…I’d rather be swapping jeans.
It’s mankind’s only fighting chance,
Designer genes, not denim pants.
We’ll stop disease and greenhouse gases,
Sequencing nucleic acids.

We’ve cracked the code, we’ve seen the light.
We’re building stuff, we’re building life…

The road ahead is clear,
Because we’re Bio-Engineers!

We’re building stuff!”