In your blog (iPhylo) an usual topic is the management and flow of information in Biology/Phylogenetics. In your opinion what are the main hurdles of this subject in science? With the ever increasing amount of generated information, databases, scientific publications, do you think that we will reach a stage where it won’t be possible to stay afloat in a sea of data?

RP- Well I think we are well past the floating stage! What I think would be cool would be a service like Google News that automatically aggregates
stories (for “stories” read “papers and data)” and summarises the Zeitgeist. This wouldn’t be overly hard to do. uBio has a service that classifies RSS feeds from journals by taxon. Imagine taking that, together with newly added GenBank sequences, and clustering the feeds by various topics. Then display that using a treemap (analogous to Marcos Weskamp’s newsmap). You could get an “at a glance” view of what is going on.

This is a bit of a half-baked idea, but it relies on things that are becoming more and more available: stable identifiers, and services that
provide data (and/or links between data). I think these are the two key things that will help us manage the flood of data. Without identifiers we will struggle to realise when we are talking about the same thing (witness the frustration with redundant links in Connotea, even though much of Connotea’s core functionality relies on stable identifiers), and services give us means to extract and manipulate data. We are slowly moving away from relying sole on web links, or thinking that merely “putting something on the web” is good enough.

Lastly, I’ve often argued that in some fields, such as phylogenetics, the days of the journal are numbered. I think what we really need is a database that can generate “reports” that function as a journal article (i.e., the contributor gets a citable paper), but where the submission process is primarily one of submitting data and protocols. Many papers in phylogenetics are of the sort “I think group x is way kewl, I grabbed gene y, used program z, and here is my tree”. Why not submit these straight to a database, instead of what we have now, which is the original data (e.g., alignments) and trees are usually lost? Databases such as TreeBASE that largely collect data after publication are sparsely populated, hence much valuable data and results are locked up in PDFs on journal web sites. Every month a weighty copy of Molecular Phylogenetics and Evolution lands on my desk, and the wasted data and trees makes me want to cry.

This is an identical question I asked Joe Felsenstein (with a twist): Lately two big projects related to phylogenetics are getting some steam: the Tree of Life and the DNA barcoding. I am more aware of the latter, due to the fact that I attended a presentation about it and through a friend that woks on the project. I think that the goals of the DNA barcoding are great for the field, wet-lab and theoretical scientists. What is your opinion about this type of project that tries to address all subjects in a big umbrella? Wouldn’t the support of a larger number of small projects be better? How would this type of projects can contribute to the information management and flow?

RP – Barcoding is the simplest of the two “big projects” — it’s a simple idea (and one that has been employed by microbiologists for decades), applied to things that we can see with the naked eye. As with all of these things, there’s been much hype, controversy, and heated debate. My own sense is that much of this is dying down, and certainly most articles in journals like Systematic Biology are investigating the performance of barcoding methods, rather than engaging in philosophical or political debate. I think barcoding will generate a lot of data, much of which will be useful, probably in ways we don’t expect. We will have standardised genetic samples taken at a known place at a known time from organisms whose identity is either known or can be inferred. I’m struggling to see why this isn’t a good thing.

My sense is that large projects don’t always scale well. You don’t get 10x more out of a project simply because it has 10x as much money. You get administrative layers, and a lot of politics. If the big project is really lots of little ones, then it may well succeed. Barcoding is really lots of little projects — everybody grabs their favourite beastie and sequences it. There aren’t any great technological or intellectual challenges. I think the real problem is when the project has big goals, articulates a grand vision, then struggles to deliver. I suspect CIPRES is an example of a large project that hasn’t delivered quite as much as it promised. My biggest fear is that the Encyclopedia of Life may end up in this category.

You have three different blogs, iPhylo, iSpecies and bioGUID, but none of them are, what I can call, Open Science. Would you consider endeavouring in this new trend and opening up your scientific activities?

RP – You missed one blog — http://semant.blogspot.com. I guess it depends what you mean by “Open Science.” These blogs are pretty open about the success and failure of various projects, and I don’t shy away from thinking out aloud. In some cases suggestions I’ve made in the blogs have been taken up and explored further by others. The Pygmybrowse widget to display large trees in a small space is a recent example.

For some time I’ve done things like post final reports on research grants online, as well as manuscripts that didn’t make it into print (these are now available in Nature Precedings).

Not all the code I write is available. Some projects, like the Taxonomic Search Engine and MyPHPBib are hosted on SourceForge, and TreeView X is also open source (code available from my site, and as part of the Debian repository). Other stuff is sometimes so experimental and fluid that it would be more trouble than it is worth to package it nicely. Some legacy code uses libraries that are not Open Source and hence can’t be distributed. That said, I’m looking at hosting current projects (such as http://bioguid.info) on Google Code. Since 2005 anything which I’ve published as first author has been Open Access, and any software or data described in those papers is Open Source.

Sometime ago, in the Nature Network’s Bioinformatics group someone was asking about good protocols in the field. Can you tell us about the Current Protocols in Bioinformatics, which is was edited by you? How content is added to it and how frequent are the updates?

RP – I’m no longer involved with Current Protocols. It was a great project to be involved with as it opened my eyes to the breadth of the field, and I got free trips to the US to meet with the other editors (I’m easily bought). Eventually I just had too many other commitments (it overlapped with being editor of Systematic Biology), so I left the board.

Content is commissioned by the editorial board, who each handle 2-3 topic areas. Updates come out every three months. If you have the print version (currently two massive ring binders), then you have the fun task of inserting new sections and removing old content.

What do you think is the importance of protocols in Bioinformatics? Isn’t bioinformatics a field more dynamic than cytogenetics and molecular biology for instance?

RP – This is an interesting point. I think there are some protocols which are widely used and are reasonably stable over time (BLAST searching, for example), and others which either are short lived (the software keeps changing, or the question being addressed by the protocol looses its relevance). In some ways Current Protocols is an old fashioned way of providing information on how to do things. A wiki may be more appropriate, certainly it could be more dynamic, and ideally free. However, one could ask whether a wiki would have acquired the same degree of authority as the Current Protocols, which could assemble a board of people with broad expertise and experience, pay contributors for their work, provide copy editing and other editorial assistance, plus give a commitment to maintain and update the publication.

Another question that is already favourite of this interview series is about computer graphic interfaces. TreeView(X) is a very important application in Phylogenetics and it is basically interface only, as the main focus is to display phylogenetic trees. Usually, a good percentage of messages sent to Paup’s discussion list are somewhat related to TreeView. What in your opinion is the importance of a graphical interface for a biologist?

RP – Well, unless you only use Lynx as your web browser, and vi to edit text, then I think the importance of the graphical interface is pretty much self evident! I think graphical user interfaces are desirable in two sets of circumstances: (1) the software is infrequently used, or (2) the user is not a “power user”. When I was at Oxford a colleague told me that he liked graphical interfaces because, after weeks in the field, the last thing he needed was to have to learn (or relearn) some garbled syntax in order to analyse his hard-won data. I’ve also seen the difference a graphical interface can make when teaching students — you want to keep the mechanics of using the software out of the way and let them focus on what they are actually doing (and why they are doing it).

Of course, graphical interfaces can get in the way as soon as you want to do something not originally thought of by the person who wrote the program. This is one reason why David Swofford’s PAUP* is a model of flexible software. There is an elegant GUI (sadly fully available on Mac Classic only, although the Windows version has some of the essential features such as the spinning cursor and the PAUP Monaco font, both things I ported across a decade ago), and the command-line interface for batch processing.

Continuing on TreeView(X), your latest version is pure wxWidgets. Why did you choose this framework to develop TreeViewX? Why not QT or Java or another GUI framework? How TreeViewX is different to the initial version of TreeView?

The short answer is that I have a lot of code written in C++, and I haven’t the time nor inclination to learn Java. When Java come along in the early ’90’s it’s big selling point was cross-platform portability. Well, it didn’t support the Mac terribly well, and I had my own C++ class libraries that worked natively on Windows and Macs, so there was no incentive to use it. Until fairly recently most Java applications I’ve seen are sluggish and/or poorly designed. When the time came to look at supporting Linux (around 2000), Qt wasn’t an option because it wasn’t free on Windows, and didn’t support the Mac. So, wxWidgets was the logical choice, although it took a while for the Mac version to be usable.

TreeView X differs from TreeView in being Open Source, and available for Windows, Mac OS X, Linux, and other forms of Unix. It lacks some of the nicer features of the original TreeView (which only runs on Windows and Macs), mainly because I haven’t had time to add them, and my own work has focussed more on web application development. Like many people who program, the software I write addresses problems I’m interested in at the time, so if my interests change, the software may languish.

In your opinion what is your greatest contribution to science. A software, a publication…

RP – I don’t think there’s any easy way to answer that. Let me respond by breaking it down into three categories. What contribution was the most fun, what was the most satisfying, and what do I think people will associate with me?

The most fun I’ve had is blogging, or to be more precise, trying to make connections between different areas such as bioinformatics, biodiversity informatics, systematics, digital libraries, and the Semantic Web. My hope is that readers of blogs like iPhylo will get something out of it, but ultimately I don’t particularly care because, hell, I’m having a ball.

In terms of satisfaction, I enjoyed working on mincut supertrees. This was a change of topic (aggregating small phylogenies into one large “supertree”), required programming algorithms (rather than, say, graphical interfaces), and was my first introduction to the world of publishing in the computer science literature (the joys of LaTeX, and the requirement of having your paper accepted for publication BEFORE you could give a talk). It was also one of the best refereeing experiences I’ve had, because one diligent referee read my source code and pointed out that what I said I was doing in the manuscript and what my code actually did were not quite the same thing (I guess you’d call this an example of “Open Science”).

But, if your question is what will people associate with me, then I think that is the work on “historical associations”, namely genes and species, hosts and parasites, and areas and phylogenies. Making the case that these are all equivalent questions, and could be approached using the same algorithm, is probably the thing I could point to as being the thing that I have have done that was “significant”.

————————————————————-

I would like to thank Dr Page for taking the time to answer my questions.

It will be ready next week and widely available for download.

Clearly it has been LONG delayed a bit. I will try soon.

—————————————————————————————

You are Biomatters’ Chief-Scientist, one of the founders of the company and the mastermind behind Geneious. What was your motivation to start the company and develop the program? How to balance these activities with your regular job as a university professor?

I would say that my main motivations for wanting to build Geneious were frustration and hope. I was frustrated at the number of molecular biologists who didn’t seem to have the software tools they needed to do their job. I was hopeful I could provide the framework for those tools, if only I had a few great programmers.

Since I was one of the only people in the department who could program, I was often asked to write scripts, or teach people how to program enough so they could do it themselves. When biologists start asking about where they can learn to program a computer, just so they can do their job you know something is wrong! Most of the things they wanted to do were reasonably straightforward. It seemed to me that these kind of basic productivity problems had been solved in other workplaces like the office — so why had they not been solved in the research laboratory? I initially wanted to build the system as an academic research project — but then realized that would fail at a number of levels. Firstly, software development isn’t science — but science now crucially relies on software development and support. Secondly, most academic programmers are not interested in (or good at) designing user interfaces, and certainly developing software is not a scientific outcome that gets recognized like publishing papers does. Thirdly, academics are quite bad at supporting software and documenting it. So it seemed to me, that for a lot of reasons, a professional software company was the best avenue to realize a software system that would dramatically improve the productivity of molecular biologists by putting bioinformatics at their fingertips. Really I just wanted to make science go faster — and with less hassle.

Geneious is a Java application? Why Java? Any special feature in the language or previous experience? Is there anything that you wish Java was capable of so you can include in Geneious?

Java is a general-purpose programming language — so you can do in Java pretty much anything you can do in software. The main reason for choosing Java is that it is very easy to write sophisticated user interfaces that run on Windows, Linux and Mac OS X. This is really important in the field of molecular biology where about 20% of the users have Mac OS X. Earlier versions of Java didn’t integrate with some aspects of different operating systems as well as they could, but with Java 5 most of this integration is quite good. Also, more and more academic developers are using Java (because it is easy, cross-platform and OO) so that also works in our favor. Every language has its pros and cons, but for our purposes Java was one of the best choices.

You are the first scientist outside North America that is interviewed here. You have worked for some years in Oxford, UK. What are the main differences of working in New Zealand in comparison to Europe? How your return to New Zealand has influenced your decision to start the company?

New Zealand is a very small country with just 4 million people. But we are a proud nation. But we have a long history of world firsts, like being the first country to give woman the right to vote (1893). A New Zealander was the first to “split the atom” (Ernest Rutherford), and a New Zealander was the first to the top of Mt Everest (Sir Edmund Hillary). We are also one of the first countries to declare ourselves a Nuclear-Free Zone (1985). So I would say that New Zealand is a place the is not afraid to try to lead the way, despite our small size. Our Prime Minister, Helen Clark, recently declared our Government’s aspiration to make New Zealand a carbon-neutral country. It is these kinds of radical ideas that we strive to make reality. So it is quite natural to try to build a unique software package in New Zealand — although we are still far from where we are aiming to be.

In your opinion what are the main differences of developing applications in academia and in the private sector?

Academics develop software for the love of it — and their first aim is to make the software work for themselves, and secondarily to work for their colleagues, as long as that doesn’t take too much effort. There are a few academics that try very hard to produce easy-to-use well-supported software, but they tend not to have the resources to do what the private sector can – for example PAUP* still doesn’t have a manual after many many years. The challenge for the private sector developing *scientific* software, is how to keep abreast of the science, which is changing so rapidly. At Biomatters we really try to have the idea that we are part of the scientific community. We have invited a number of scientist to work with us on various projects and we find that in this way we learn a lot more about what is needed, and how best we can fit into the academic ecology. Our goal is a happy marriage where academic programmers can get on with developing great new algorithms, and Geneious can provide the interoperability, the user interface and the support.

What would you teach to an “academic developer”? What developers in academia lack when compared to developers working in companies?

I don’t think it is a question of what academic developers lack in terms of skills — I think it is more a question of motivation. Aside from Geneious I produce academic software (like BEAST). What motivates me to answer 100 technical email queries every month about the BEAST software? My main passion is science, not supporting software. So by creating a successful software package, I have created a heavy burden of support. I would rather that somebody else supported all the simple questions about user interface, file conversion, installation, et cetera… So in some sense, what I am hoping to “teach” academic developers is that maybe there is another way to do things. It we had one framework that made interoperability and user interface more seamless and standard, each academic could focus on the science, the new algorithms and the part of their job that they find really exciting — the discovery.

Another question that is already favourite of this interview series is about computer graphic interfaces. Geneious has a very nice GUI, but in my opinion still lacks some simple features. Can you tell us how the program’s GUI was first designed and what kind of user input is considered in the development? Which is more important: “what the user want” or “what we want”?

Most users didn’t say they wanted the mouse, or the iPod before the visionary told them they wanted it. So I do think there is room for a lot of innovation in software development. But of course, there are definitely things that users know that they need right now. Balancing these two things is the real trick. Biomatters is very serious about listening to our users and we have an active forum and a small group of power users that have been quite helpful in the past. Of course there is always more to do. We also think that we have introduced a few nice new ideas in our software and there are more on the way. Specifically we are interested in developing the concept of “live” research publications, where arrival of relevant new data automatically updates the published research in that area. As you can imagine, this is a long term goal. But I think you need these visions to aim for, if you are going to make real progress.

For the young bioinformatician: should s/he focus on publications, skills, languages? Or a balance of all? What is your advice?

A young bioinformatician should focus on the science. What are the outstanding questions that your skills enable you to take a step towards answering? Once you focus on the science the publications, skills and languages will come.

I used to program ten different languages very well. But 95% of my programming these days is done in Java (because I am usually building on stuff I have done previously). PERL is still quite important in bioinformatics — but I think that is because there is still a lot of need to write one-off scripts. Will we get to a point where there is more sophistication? I think that bioinformatics has to become a field where people without programming skills can contribute substantially. I would argue that all of the programmers in bioinformatics should be working very hard to program themselves out of their jobs (and into more satisfying jobs). Lets build software systems that take this field to the next level. That is what we am trying to do (slowly but surely) with Geneious. We want to create a system that ultimately makes the programmers redundant — at least for the menial tasks — so that any molecular biologist can construct complicated queries and analyses in an intuitive graphical manner. Then the programmers can focus on the hard questions like new algorithms and methods of analysis.

What in your opinion is your most important software?

I don’t really compare my software projects, because they are so different. I guess this is for other people to decide. But judging from you line of questioning, Geneious is currently winning

————————————

I would like to thank Alexei for answering my questions.

—————————————————-

My initial questions are about Open Science. What are the main benefits of opening your whole scientific life to the public? And the drawbacks of doing this? How is the best way to manage a lab under Open Science?

RR: There are practical consequences and philosophical consequences to doing open science the way we are. So far the practical consequences have been good – I get helpful suggestions from people I’ve never met. And I can tell people who wonder what we’re doing to take a look at the stuff on our web pages. I’d love to also be able to say that our open approach has attracted collaborators and new lab members, but that hasn’t happened yet. Another big practical consequence is that writing blog posts about the research I’m working on helps me think clearly about it. So whenever I’m feeling muddled I post the problem on my blog.

Philosophically it just feels right. Science is fundamentally a collaborative endeavour; we all build on what others have done. Just as the web gives us new and better ways to publish our findings in peer-reviewed journals, it lets us find out communicate about our work in progress. Furthermore, the taxpayer pays for our work, and deserves to be able to see what’s going on. So I like to think of random people stumbling onto my blog and thinking “Hmm, so that’s what a scientist really does.”.

Also, openness is much easier and more comfortable than secrecy. In my teaching (mostly first-year biology) I give only open-book exams, and it’s much easier for me than worrying about cheaters.

I can see on your webpage that you have grant applications, results and even future plans listed. Do you fear having your ideas and plans “stolen”? You might have some data and results that end up on the webpage/blog only after the publication or everything is uploaded as soon as it out of the oven?

RR: I blog about the day-to-day research we’re doing, and I post our grant proposals as soon as they’re submitted. So everything is open from the start.

I don’t worry much about people stealing our ideas, but it’s not a big risk because nobody else is working on what we’re working on. And most of the people doing related work seem to think I’m more-or-less misguided in my interests and approaches. I try not to post exactly the same figures I’m putting into a manuscript (only because I fear the journal editors might get cranky), but otherwise my goal is to give anyone who’s interested an accurate picture of what we’re doing.

The visibility of scientific blogs has increased considerably in the past years. Yours is up for almost an year now (at least your archives go back to August 2006) and it seems that you support that your whole group of students and post-docs do the same. What have influenced you to start recording your daily routine on a website? Does it help the interaction in the lab since everyone has its own personal page?

RR: I started blogging about research in response to a post by Pedro Beltrao (http://pbeltrao.blogspot.com/2006/07/opening-up-scientific-process-during.html) about the desirability of opening up the scientific process (I found it through Tangled Bank #59). As soon as I read it I realized that I wanted to have a blog about my lab’s research.

After a couple of weeks I told everyone in my lab that I wanted them to have blogs too. I love reading their blogs – they have such thoughtful things to say about their research. They don’t always post as often as I’d like them to; they say it’s because they feel there’s a high standard that they can only sometimes reach. But I’d like them to at least scribble a few lines every day.

From your blog it is easy to see some frustration with bioinformatics. How important is bioinformatics in your research and what the major problems that you have with it?

RR: It’s a collection of important tools and resources, and I love what I can do. And I enjoy learning new things, like babytalk Perl and standalone BLAST. But Unix has always driven me crazy – I keep hoping to attract a Unix/Perl geek to the lab, but no luck so far.

My bioinformatics frustrations arise in two ways. First, it’s only part of what we do, so I’m not doing it nearly enough to get really good at it. If bioinformatics was the main focus of our work I would invest more of my brain in improving my skills. But we’re primarily an experimental lab (we get our hands wet). And each time I go back to it, NCBI or TIGR has changed their interfaces.

The other source of frustration is that we’re always trying to do new things, that nobody has ever done before, Sometimes we can find existing tools that can do these things, but sometimes we have to write our own and we don’t really know what we’re doing. For a while we were lucky enough to have Mario Pineda-Krch working with us, but he’s gone on to better things (see http://pineda-krch.com/).

Another question that is already favourite of this interview series is about computer graphic interfaces. From what I could see your lab generates most of its scripts in Perl, but some work is done with graphical interfaces. What do you and your group look for in a program and its graphic interface? What are the main drawbacks of the command line interface that could be easily mitigated by a GUI?

RR: By graphic interfaces do you mean things like the web interfaces for BLAST? Basically we’ll take whatever we can get. If the resource has a nice point-and-click interface, great. If we have to submit a string of instructions to get what we need, we can cope. If we’d need to write some software ourselves, we usually move on to a different problem unless we’re highly motivated and the problem looks simple. I hate the frustration of not getting line commands exactly right, so much prefer menu-driven interfaces.

The only good interface we’ve generated was done by Mario, working on the fabulous hotspot paradox.

How do you see the “publish or perish” in science, with the increasing number of retracted papers, fake results, rushed publications and sometimes publication of extremely similar results in different papers and journals? Being an advocate of Open Science, do you think this new movement can help alleviate some of the known problems of the “publish or perish” culture?

RR: There’s always been a lot of bad science mixed in with the good; I’m not convinced that the proportion has gotten higher as the total number of papers has increased.

Do you have an advice for the young scientist? And, as you are the first woman interviewed here, any special advice for the girls that are considering a scientific career?

RR: Work by women is unconsciously but systematically undervalued by everyone (including the women who do it), so don’t go into science for the glory. But if you love doing research it’s wonderful. I’m still delighted that I can get paid to do this.

What in your opinion is your most important publication?

RR: That’s a hard one. I don’t get much recognition for what I think is important, mostly because my conclusions don’t fit what others want to hear.

——————–

I would like to thank again for Dr Redfield’s answers. I am planning some changes to SciView soon.

If you want to follow the RRResearch blog, here is the link.

With this interview the first cycle, which I might call inaugural cycle, is complete. I would like to thank everyone for their support and if you have any comments or ideas just drop me a note. Probably in the coming months, SciView will have a home for itself and added features.

——————————————————-

You have started your scientific career with conventional cytogenetics and doing analysis with optical microscopes. What have changed since you started, with the advances in technology. new methods,new equipments? Do you still believe that basic cytogenetics analysis is still strong?

JS: The big advance was application of CCD cameras and the computer interface to the scopes. The basic analytical platform is still 70% determined by the human brain as computers are unable to interpret the biological and technical variables of banded metaphase preparations correctly.

The other big advance was the development of FISH methods and the parallel availability of the toolbox of probes and techniques created by the human genome project. This has enhanced rather than replaced my area so that basic cytogenetics has wider applications now, than in 1973 when I first started looking at human chromosomes.

How important is bioinformatics for human genetics and oncology? With the ever increasing presence of computers in any research, how do you see the introduction of bioinformatics in your field?

JS: Bioinformatics in its various forms has played a vital role in bringing all the “omics areas” back to the bench and it is an essential ingredient in most new hypothesis driven lab projects in genetics and oncology. Like most scientists brought up on wet bench research I have a high level of skepticism of in silico projects that do not touch down to test simulations and speculations by experimental methods. This has been a weakness on some grant proposal that I have reviewed in recent years. I my field I have covered past and present issues above. In the future I see bioinformatics helping understand the 3 D structure of genomes (in cancer and normal nuclei) being essential.

Another question that is already favourite of this interview series is about computer graphic interfaces. Clearly for your type of research the graphic interface is a must have, especially in cytogenetics and microarray applications. Based on your experience would you say that the adoption of an application in your lab could be determined by the quality of its graphic interface? What do you and your group look for in
a program and its graphic interface?

JS: Absolutely agree. The graphic interface has to mirror the scientific thought processes that we use when we conceptualize the data. Our experience with the Agilent array platform has been great and it has helped up move faster in integrating different genomic findings. Apart from the overall output being elegant and easy to understand it should be transparent and adaptable to our own needs. Also in more practical terms – working on Apple and PC would help a lot. Plus being able to take on your own laptop to play with whilst waiting at airports (I have done that a lot this month!)

What is your take on the main differences of oncology research between US and Canada? It seems that Canada even with a smaller number of research institutes have the lead in different fields of cancer research.

JS: In human cancer Canada has a stronger programs in signaling cancer research and cancer stem cell approaches. This relates to successes in the 1960s (stem cells) and then in the 1980s (signaling). Cancer genetics and genomics is stronger in the US, as is translational cancer research. Our funding system favours individual PI driven projects that tend to be more basic. Our genome project was poorly supported in the late 1990s. We need to have NIH SPORE grants for clinical cancer research projects as these encourage team approaches of scientists and clinicians to work together. Canada is excellent in model organism (mouse, flies, yeast) approaches to cancer research. The weakness in applied cancer research in Canada is that the bench to bedside work is used to raise money from donors but there are few Canadian success stories to talk about.

You are still in Brazil doing a sabbatical. From what you have seen of Brazilian research, how do you compare the science in a country under development, with a young science sometimes underfunded, and the science in major centres like US, Canada and Europe? What is the main advice for a young scientist who wants to come to a major centre?

JS: This is a big topic and I would prefer to talk to you about it over a beer! I think any serious Brazilian scientist has to do a postdoc in NAmerica or EU. If they love Brazil more than they love research they should develop some serious political connections with FAPESP before they leave to do their postdoc. This will help them when they come back. However it will always be frustrating to do good research in Brazil. Too many PIs here are not serious about research.

I asked this question to Joe Felsenstein and Brian Golding: How do you see the “publish or perish” in science, with the increasing number of retracted papers, fake results, rushed publications and sometimes publication of extremely similar results in different papers and journals? What are your views, especially in a very competitive field as oncology?

JS: As long as grant panels judge researchers by productivity this situation will not change. To address this area I support the idea of declaring the identity of the reviewer of your grants or your submitted paper. I think more thought would go into the review process and to the quality of submitted papers if the process was more transparent. Some journal are doing this now. In the oncology field I would like some of the journal to have part of the abstract devoted to the translational component of the study.

Do you have an advice for the young researcher in human genetics? Should s/he focus on a determined trend or be a generalist in the field?

I think this entirely depends on how the young researchers own neurons are wired. Some people function better as generalists and others need focus. There are of course intermediate versions. I would advise someone starting out to be opportunistic and to look very closely at emerging trends for the choice of a postdoc for example.

What in your opinion is your most important publication?

AS STUDENT:
1. Godbout R, Dryja TP, Squire JA, Gallie BL, and Phillips RA. Somatic inactivation of genes on chromosome 13 is a common event in retinoblastoma. Nature. 304: 451-453, 1983.
AS PI:
2. Kolomietz E, Al-Maghrabi J, Brennan S, Karaskova J, Minkin S, Lipton J, and Squire JA. Primary chromosomal rearrangements of leukemia are frequently accompanied by extensive submicroscopic
deletions and may lead to altered prognosis. Blood. 97(11): 3581-3588, 2001.
——————————————-

I would like to thank Joe Felsenstein, Brian Golding and Jeremy Squire for taking their time to answer my questions. SciView will be back (hopefully) soon.

————————————————————————–
In your opinion is there any difference about the Bioinformatics research done in Canada and in the US? The number of companies, universities and institutes in Canada is evidently smaller than in the US, would that be an obstacle for the development of Canadian Bioinformatics?

BG: Canada has often had a “branch-plant” economy for many companies. So little primary research is done for the major pharmaceutical companies here. Because genomics and pharmaceutical research is expensive, the research here is probably less per capita and of a different nature (for example more testing). I don’t think that this is an obstacle, it simply means that the research for bioinformatics is mainly centered in universities and mainly in a few concentrated areas.

I asked this question to Joe Felsenstein: How do you see the “publish or perish” in science, with the increasing number of retracted papers, fake results, rushed publications and sometimes publication of extremely similar results in different papers and journals? What are your views?

BG:Yes, the paradigm to publish is getting worse than when I was younger. Indeed, I would not have had enough papers to get a job. It has also lead to many more authors per paper than in the remote past when I was younger. The increased pressure has an obvious consequence that you note above. However, this is not limited to research. The need for “accountability” is a buzz word in politics today and any government venture has to be “made accountable” (since for some reason it is perceived not to have been in the past). Research with government dollars is painted with the same brush. Unfortunately, for much of science, increased accountability in an already very open system, simply leads to more paperwork and less productivity (less accountability).

You have been an advocate of free/open source software. In your opinion what is the importance of this type of software for the Bioinformatics research? Do you agree of academic researcher to pursue copyright/patents of applications that are developed using public money?

BG: I do not agree with this though I find my opinion to be in the minority. I have always found open source software to be excellent work (people base their reputations rather than their cash on the software). You do require some willingness to accept strange bugs but people that pay for software seem to be willing to accept these same bugs that they are intolerant of in open source. I think that many are living by the adage that “you get what you pay for”. I think that there are many exceptions to this and a community based software project is one of these exceptions.

Another question that will become a favourite of this interview series is about computer graphic interfaces. There are many scientific software that does not have interfaces. Based on that, do you think that the lack of GUI would be an obstacle for the adoption of a program in research?

BG: Having a command line option for a program is a requirement for me. I find the command line to be the natural place to “get things done” in LINUX. It provides a way to use the power of the shell, the operating system and other programs to easily interface with whatever other program I may be running. Without it, you are limited to the interface that was created by the programmers.

How do you see that some bioinformatics companies create proprietary packages that are mere wrappers of open source academic software? Geneious, CLC Workbench, etc are excellent GUIs but the main engine is/was developed using academic research.

BG: This is common practice. They are selling value added software. So are the LINUX distribution packagers. They sell LINUX and a package of software that goes with it, each itself usually free but checked to make sure they all play nice together. The community is one where they usually also make available a lesser package that contians the essentials for free. If it is important to people to have a package that they need not worry about piecing together then they can pay for it. I am however, surprised to find any people that have paid big bucks for packages that I piece together for free often with a more powerful interface. Difference is, I am willing to spend some time to try to learn.

EvolDir is a great service for the scientific community (I found my first post-doc through it). I know that it is run using your own software, based on scripts and homebrew code. Why did you decide to rely on your own scripts instead of using Mailman or Majordomo? How EvolDir was started?

BG: Mailman and Majordomo did not exist when I started the EvolDir. The EvolDir was started simply as a file that contained a list of emails that I knew (this was before the days of web browsers or even gopher browsers and the only way to get an email was to ask someone for it (by voice only). A friend saw it and asked if he could get a copy. The list changed shortly after so I thought that I should send him another, updated copy. Another person asked and so on; strictly via word-of-mouth (literally in those days). This is why it is called a Dir (directory; a list of names and emails). Notices of general interest to people came later.

What would be your advice for someone that starting in the field? You came from the wetlab and probably have used punch cards in the early days of informatics. Is the life of the young bioinformatician easier than in the old days?

BG: A bioinformatician today can do things that I could not imagine when I started as an undergraduate. This morning I BLAST’ed over 5000 proteins against more than 500 complete genomes (in about 20 minutes). When one thinks of all that this entails it is staggering. The computers are so fast today that the trick to getting things done is no longer how quickly you can contrive to get back into the queue for those few precious cpu seconds. Now most people, except for some specific programs or gaming, cannot keep a computer busy … hence the creation of the grid to give this idleness to others that can. The world is a much different place, biology is much different and what it even more exciting, I would surely anticipate (given the advances that one sees today) that in another ten/fifteen years, the world and biology will be different anew.

———————————————————————————-

How do you see the “publish or perish” in science, with the increasing number of retracted papers, fake results, rushed publications and sometimes publication of extremely similar results in different papers and journals?

JF: As you imply, “publish or perish” is a Bad Thing, which leads to all these negative results. We should all be given a lifetime stipend with freely available grant funds, in hopes that before retirement we will produce at least one passable paper.

Phylip is a great piece of software, in most cases better than PAUP and other similar applications. Can this be linked to the fact that it is open source and freely available? Have this helped to keep the software updated and portable to any system?

JF: You will find that opinions vary greatly as to which is better, PHYLIP or PAUP*. More people find PAUP* better than find PHYLIP better. So your premise may be wrong, which leaves us without a need to find the reason. The availability of the source code allows people to suggest changes that increase robustness and portability. But another reason is simply that I tend to write (and urge my programmers to write) in a “paranoid” style of C which deliberately avoids local features of one operating system or features of C that are sometimes not implemented correctly. One correction: PHYLIP is *not* Open Source. It is copyrighted to the University of Washington, and resale of it is potentially subject to a license fee.

Actually PHYLIP is not the result of a large community of users. Mostly it’s me and my student programmers, with some important pieces added by others. So it is not an example of the strength of the Open Source or GNU approach.

Recently, I have focused on improving my interface design skills, trying to make the bioinformatics software-user interaction more friendly. Do you think that due to the fact that Phylip does not have a graphical interface was an obstacle to make it more widespread?

JF: People often ask me why it doesn’t have a GUI. They also frequently request the opposite: a command-line interface with all options invoked from the command line. The problem is not having some GUI, it is having a good, intuitive GUI. That is a matter of design (Intelligent Design, I guess). I once assigned a very good programmer to make a demonstration GUI for one of my programs, mimicking the menu that we already had (and actually then doing nothing, just showing what it would look like). The result was so totally boring that it alerted me that the issue was not a GUI but what kind of GUI.

How do you see that some bioinformatics companies create proprietary packages that are mere wrappers of open source academic software? And what about academic software that is sold? Most of them are funded by public agencies and tax money.

JF: In the USA in the 1970s the standard assumption was that publicly funded research should not be exploited commercially without a royalty coming back to the funding agency and the researcher. Under the Reagan administration in the 1980s the view changed to being that the duty of the government was to subsidize the private sector by having their results used without compensation. The very same administrators who, in the 1970s, checked to see that you had not secretly privatized the products of research were now assigned to make sure that these results did get used privately (without compensation). So it is hard to say what the legal status of research products such as software is. (There is an older tradition that books written by faculty members at universities can result in royalties to the author, without the university getting any payment).

One problem with Open Source (and with GNU licenses) is that they permit and encourage commercial exploitation without compensating anyone. They love this but I would rather see some of the money made getting back to the original software authors. That is one reason we aren’t licensed under Open Source or under any of the GNU licenses. I also have questions about whether these licenses permit the original author to say yea or nay to a proposed added feature — I would prefer to maintain that degree of control.

Your seminal work in 1981, introduced the maximum likelihood method for the calculation of phylogenetic trees. I do believe that a scientist’s work has to be evaluated in its entirety, but that Journal of Molecular Evolution publication has influenced thousands of researchers for almost 26 years. Would you consider it your most important publication?

JF: The maximum likelihood method for phylogenies was introduced by Edwards and Cavalli-Sforza in 1964, and for molecular sequences it was first done by the great statistician Jerzy Neyman in 1971. My 1981 paper did cite Neyman (it didn’t credit Edwards and Cavalli-Sforza but at least I have cited their pioneering work on this many times before and since). What I did that was new was to make it practical to do multi-species cases by introducing the pruning algorithm. I also promoted and explained ML for years. In the 1980s I gave many talks at universities. I usually had to explain what likelihood was because the statistics courses that biologists took didn’t bother to teach them this. I also wrote a number of review articles and made programs available to do ML. Of my papers, the 1981 ML paper is the third most cited so far (after the 1985 bootstrap paper, a paper in 1989 in Cladistics describing PHYLIP, and just ahead of my 1985 paper on phylogenetic comparative methods). Basically they are all part of the process of promoting a statistical modelling approach to phylogenies, so it is hard to label one as more important. Possibly my most important paper in the long run will be my 1978 paper on a model of macroevolution, which has had no citations in the past 22 years.

Sometime ago I attended a Hennig Society meeting, mainly because it was organized by the same institute where I was doing my PhD and because I was curious to meet the researchers behind some of the papers I was reading at the time. Later when I was doing part of my work at the National Museum of Natural History, I attended a lecture by James Farris and what happened in that presentation made me coin (for my own usage) the term “cult” for cladists and cladistics. What is your take on the usual “cult”-followers narrow-minded scientific views, where no method is better than them and sometimes even alignment software is demonized.

JF: Now you have some sense of what the atmosphere was like in morphological systematics in the early 1980s. At that time most young systematists had joined the Hennig Society in a fit of enthusiasm for phylogenetic systematics, and work coming from any other intellectual source stood a good chance of being rejected out of hand as “not being science”. My own guess is that this level of excessive zeal was a reaction to the inertia in systematics, as all the main professorships and curatorships then were in the hands of followers of Mayr and Simpson, the “evolutionary systematists”, who would not make their methods explicit. The way to intimidate them was to up the temperature of the debate. This attracted people with certain kinds of personalities.

You aren’t the first person to analogize the behavior of this group to an evangelical religion. The Hennig Society types have always wanted to be seen as having a key to understanding everything phylogenetic, a key that everyone else lacked. In the longer run, their behavior alienated their own base, and there were internal conflicts. Most young systematists deserted the Hennig Society and started coming to the Evolution Meetings, where the discussion is fairly normal in tone, personal attacks are discouraged, and there is a lot of good-natured discussion among people with different views.

(I don’t think the issue is whether or not one uses cladistic approaches to classification, as most of the reasonable young systematists at the Evolution meetings take that position).

What interests me is that the hard core of the Hennig Society has not mellowed or compromised. I was expecting that they would, but they have not only remained as ferocious, but continued to hold views very distinct from everyone else’s. They have circled the wagons and are still holding out, though they must be getting a bit lonely by now.

Lately two big projects related to phylogenetics are getting some steam: the Tree of Life and the DNA barcoding. I am more aware of the latter, due to the fact that I attended a presentation about it and through a friend that woks on the project. I think that the goals of the
DNA barcoding are great for the field, wet-lab and theoretical scientists. What is your opinion about this type of project that tries to address all subjects in a big umbrella? Wouldn’t the support of a larger number of small projects be better?

JF: I don’t really object to big projects as long as they don’t suck up the money for everything else. I was on the Genbank Advisory Committee some years ago, and I think the DNA and protein databases and Pubmed are examples of important and necessary large-scale projects. So although I worry about money for small projects being soaked up by the big ones, I don’t object to the latter in principle. In Canada they used to have a system of giving out small amounts of money to everybody. They are moving away from that now, but I don’t think it should disappear entirely.

You have in Phylip’s grant webpage a “no thanks” section listing everyone that refused supporting the program along the years. In my opinion this is a bold statement and not very common in the scientific environment (at least not online). What would be your advice for the young scientist that is searching for financial support for his/her research? Should s/he be vocal when a strong application is rejected?

JF: Generally, no. I could do this because I felt confident enough in my reputation. It was fun. These granting agencies can say all sorts of fatuous things in their evaluations and they are never called on this. A web page is editor-free publication so I had fun with that. I had a hope that it might cause grant reviewers to think twice about getting themselves on that list by coming up with arbitrary and ill-thought-out objections. (Based on our lab’s experience since then, alas, this isn’t working). But for a vulnerable young researcher I would say no, don’t succumb to that temptation. You may get a reputation as a sorehead. I know a few young scientists who are quite combative about any negative evaluation, and they do get a reputation as someone you don’t want to deal with. However in these days of increasing difficulty in finding funds, it is tough for newcomers as the agencies are looking for any reason to say no.