Software For Life

Blogs and Twitter at Scientific Conferences

2009-06-04T19:53:00.000-07:00

There's been a mini-furor over how a recent Cold Spring Harbor Conference triggered a complaint by the magazine Genome Technology about how their journalists were under restrictions that bloggers and twitterers weren't. Here's another blog entry to catch you up.

My opinion is that conferences should ask each speaker if they'd be willing to be reported on by anybody, print or on-line. If they say yes then precede their talk with an announcement of that fact or print it in the agenda. I find it hard to believe that any scientist will lose out with more publicity. There's always the possibility that the reporting will be wrong, but I've read lots of misleading print media reporting on scientific topics, so that problem already exists. The bloggers present at scientific meetings are most likely to be scientists themselves so I suspect they'll do a good job. If they don't then they'll get corrected by comments or by the presenters themselves afterwards.

Cold Spring Harbor's decision is to force bloggers to get permission in advance just like journalists before writing about a speaker. I think this is a mistake. Instead of restricting bloggers just like journalists, they should make journalists as free as bloggers.

What we're most excited about

2009-05-20T18:33:00.000-07:00

Personalized medicine and next-gen sequencing are the most exciting, and most hyped, things happening in the field. The presence of both was felt at the conference because both require lots of informatics/IT support. As realists, we want to dampen our excitement somewhat because the flood of data (as you'll read about below) doesn't always lead to more knowledge. As optimists, however, we believe that the promise of personalized medicine and the potential illumination yielded by next-gen sequencing are real, not hype alone.

Thoughts on the Future of Personalized Medicine

2009-05-20T17:00:00.000-07:00

There was a panel discussion at Bio-IT World that focused on personalized medicine. It included people from Navigenics, Knome, the 1000 Genomes Project, and Complete Genomics, among others. It's clear that genotype data, and soon complete genomic sequence data, will be available for a large number of people. However, you do have to make the distinction between genetic risk data that is useful for the scientific community for research purposes, and that which is useful for a particular individual. There's no doubt that the results of genome-wide association studies (GWAS) are providing a lot of interesting data for researchers and informaticians, but the leap to say that this data is useful for the individual is less obvious. Most genetic risk factors are relatively weak by themselves, and often provide a prediction of increased risk that is very modest. Additionally, such risk factors as APOE for Alzheimer's don't come with any preventative treatment options so it's unclear what benefit they have for the individual.

There's even discussion that the whole rationale for genome-wide association studies using single nucleotide polymorphisms (SNPs) might be somewhat flawed. The effect of rare SNPs, which are not easily assessed using the current size of GWAS studies, and of copy number variants, might be much larger than previously thought. In short, I think there will likely be a lot of benefit for individuals down the road but it's clear there is a lot to learn before that becomes a reality.

What Next-Gen Sequencing Data Can and Can't Do

2009-05-20T16:00:00.000-07:00

I was fortunate enough to attend the American Association of Cancer Researchers (AACR) meeting in Denver the week before Bio-IT World. Bio-IT World was filled with talks and discussions about how technologies such as next-generation sequencing, which have the potential to generate huge amounts of data, are requiring changes in IT and informatics. But a theme at AACR, from several speakers, was that data itself does not solve the problems of understanding the underlying biology of disease. And understanding the biology is how new therapeutics and diagnostics are going to come about.

Clifford Reid, CEO of Complete Genomics, said in a keynote at Bio-IT World that their data would allow us to understand cancer in five years. I think if he'd made that statement at AACR he would have a hard time. Cancer researchers are glad for this kind of data but they are also aware that it's not worth much if you can't use it to gain biological knowledge. Even though there are IT and informatics challenges from these new technologies, the biggest challenges are going to be how to get this data into a form that it can be assessed by those that have expertise in the biology of disease.

Technology provides flexibility

2009-05-19T12:33:00.000-07:00

Gathering clinical trial information rapidly, and acting on it, allows research to react flexibly to its own outcomes, saving time and money. Technology allows researchers to view genomic data in a systematic way (as a system), revealing new insights. The BIO IT World conference attendees are a biased crowd, believers that software can help, witnessed by the provocatively titled session about how technology "...will Help Clinical Investigators Reduce Their Paperwork and Get Back to their Real Work."

Integration supports better understanding, faster results

2009-05-16T20:51:00.000-07:00

Georgetown's Lombardi Cancer Center is attempting to bring together disparate systems so they can better make sense of their information and treat patients. NCI's Ken Beutow spoke of cycling between clinical care, research, and back. Integration is more than ETL, it requires understanding the business purpose of different systems or databases, and finding the right way to integrate information to provide the right pathways and answers.

Share, baby, share

2009-05-15T23:14:00.000-07:00

And share "freely and generously," as Benjamin Franklin put it - that idea was echoed by many. Notionally, it's simple; technically...a bit more complicated. The standards debate will continue. That said, the appeal of the benefits that standards provide - access to information - can be outweighed by their implementation cost. Standards can help information be shared and used in ways its originators hadn't imagined...as Eric Schadt put it, "We're not going to make progress by publishing papers" - understanding results, and having access to data, spurs progress.

Does the format of the information matter, or does access to the information matter? Is a large infrastructure required to support sharing? Somewhere between an emailed spreadsheet and a complex interoperable infrastructure lies a way to get information in a meaningful way, easily. That's all we need to do to speed the pace of progress, huh? Guess that's why most of us are here.

Cloud-y days ahead

2009-05-14T12:35:00.000-07:00

The amount of data we're generating continues to grow immensely. For every time we heard "next-gen sequencing" from one person, we'd hear "cloud" from another. Storing information on the cloud may support it all - integration, flexibility, and sharing. Chris Dagdigian's exploration of the benefits and downsides to storing information on the cloud was thrilling, but reminded us that one of the critical components of sharing data is the access route itself - i/o issues may slow adoption. The imbalance is revealed by the fact that it can cost pennies to store information on the cloud, but hundreds of dollars to get the information there.

Revealing information in data remains a challenge

2009-05-13T16:26:00.000-07:00

The massive amounts of data experiments produce can create management challenges - where to store it (cloud? spreadsheets? that old server?), how to determine whether it's worth keeping around in the long term, and figuring out how (and whether) to analyze it (how can we gather just the pieces that are useful to us?).

Where to store it will be a continuing question - is the cloud a buzzword or viable solution? A common refrain was that it's cheaper to re-run some tests than it is to keep test results. As for analyzing results, or being able to winnow the data down to a subset worth analyzing, we see the challenge of making sense of data - connecting the dots - as one of the most vexing we face. As technologists, we presume software can help, but know that if software's going to help, it's got to be smart, flexible, and easy to use. Unlocking the answers in data is key; being able to do it before we drown in data is the hard part.

Can you guess the data management winners?

2009-05-12T06:29:00.000-07:00

5AM conducted a survey of BIO-IT World attendees, and preliminary results show that while nearly all survey respondents declare "the data flood" as one their largest challenges, the most common tools being used to manage data are email and spreadsheets. We'll be publishing survey results soon - please click here if you'd like to receive them (we will not share your personal information).

We all understand why these old standbys remain the dominant data management tools - they're universal, and they're easy to use. With the flood of data we're now producing, however, they're reaching their limits to assist. At 5AM, we believe that simple, open-source tools can take researchers, pharmaceuticals, labs, tissue banks, clinicians, and other life science professionals to the next level. Contact us if you'd like to talk about how we can work with you to provide the solution that meets your needs. We welcome your comments on these topics and look forward to sharing ideas together at BIO-IT World in 2010! -5AM