Sign in or Join FriendFeed
FriendFeed is the easiest way to share online. Learn more »

Adam Kraut › Likes

Lars Juhl Jensen
Commentary: The GPU computing fallacy - http://larsjuhljensen.wordpress.com/2011...
Modern graphics processors (GPUs) deliver considerably more brute force computational power than traditional processors (CPUs). With NVIDIA’s launch of CUDA, general purpose GPU computing has become greatly simplified, and many research groups around the world have consequently experimented with how one can harvest the power of GPUs to speed up scientific computing. This is also [...] - Lars Juhl Jensen
Brad Chapman
Next generation sequencing information management and analysis system for Galaxy - http://bcbio.wordpress.com/2011...
Pierre Lindenbaum
Translating a DNA to a Protein using server-side javascript and C: my notebook - http://plindenbaum.blogspot.com/2010...
Bosco Ho
Chris Miller
"Perl – The only language that looks the same before and after RSA encryption." -- Keith Bostic
Unreadable? - Egon Willighagen
:p pff, I will not be forced to indent my code ! :) - Pedro Beltrao
And Perl is the only language that has both "white" and "black" magic. :) - Pawel Szczesny
Roderic Page
Viewing scientific articles on the iPad: cloning the Nature.com iPhone app using jQuery Mobile - http://iphylo.blogspot.com/2010...
Deepak Singh
RT @mza: New #aws #hpc tutorial video: Building a High Performance Cluster on #ec2: 64 cores, 800Gb storage, 10 minutes: http://dicjtockkg63v.cloudfront.net/hpc-vid...
Deepak Singh
Are expectations being set properly? - http://scalability.org/?p=2816
So true - Deepak Singh
Rajarshi Guha
Supercomputer sets protein-folding record : Nature News - http://www.nature.com/news...
What a story! Academia --> hedge fund --> ridiculous wealth --> self-funded academic research - Andrew Su
And I remain completely in awe of DE Shaw - Deepak Singh
Wow, what would you do in science if you had that amount of money available? - joergkurtwegner
Shaw is not doing this for commercial reasons. He has the money, which gives him the luxury to try interesting problems that don't make commercial sense. Although as Vijay Pande says, even Shaw acknowledges that Anton only goes so far - Deepak Singh
Anyone done an Anton vs. Folding@Home comparison? - Matthew Todd
Wonder how long that'll hold up with Blue Waters around the corner. http://gladiator.ncsa.illinois.edu/vidcast... - Christopher Fields
Rajarshi Guha
Deepak Singh
Back to the future? - http://mndoci.com/2010...
Mmmmmmaybe. I like a one-stop shop myself. - Bill Hooker
I also think that some of the mechanisms for notifications and discovery have evolved to a point where this is possible. Not to a point that it's generally usable, but for the geeks, it might just work, at least as a communication and discovery channel. The one thing we still lack is a good identity system. The best one out there is Facebook, but in my circles, there is enough resistance to it. Which leaves OpenID, but, alas, it's not quite as well adopted as one would have oped - Deepak Singh
to followup what Deepak said, if it's good enough for the geeks, it's good enough. Does it really need to be Facebook-scale before it's useful? Maybe not having millions of people and the associated spam issues could actually be a feature? - Mr. Gunn
Pavolga
More dates/locations open for the #Cloud for Life Sciences training with the Bioteam's Adam Kraut & Chris Dagdigian): http://www.healthtech.com/cloud #AWS
Roderic Page
CouchDB, Mendeley, and what I really want in an iPad article viewer - http://iphylo.blogspot.com/2010...
Nice post! - Egon Willighagen
Rajarshi Guha
Pig and Cheminformatics - http://blog.rguha.net/?p=748
Blog is not accessible, got in feed reader. - Abhishek Tiwari
Blog down for me too (9:18am UK) - Noel O'Boyle
Worked for me two hours ago... (~9am CEST) - Egon Willighagen
Aargh. Back up now - Rajarshi Guha
Deepak Singh
Chris Lasher
Why working at home is both awesome and horrible - The Oatmeal - http://theoatmeal.com/comics...
[Found via Khader Shameer] - Chris Lasher
I have 1h commute and the boss does not mind that I stay home once in a while but more than a 1 day per week is not very good for you :) . It works well when I have a very specific thing I can focus on, then I get a lot more done at home. If I am playing around with some ideas then I just get too easily distracted at home. - Pedro Beltrao
Pierre Lindenbaum
Mwncbi, a mediawiki extension loading asynchronously some records from the NCBI - http://plindenbaum.blogspot.com/2010...
frickin' cool... - Andrew Su
Pierre Lindenbaum
A Tiny Genome Browser (XHTML/Javascript/json/svg) - http://plindenbaum.blogspot.com/2010...
Jan Aerts
"Intro to Clojure" video - http://parleys.com/#sl=13
Donnie Berkholz
Using a conformation-dependent stereochemical library improves crystallographic refinement of proteins. 2010; Acta Crystallographica Section D - Wiley InterScience - http://www3.interscience.wiley.com/journal...
"The major macromolecular crystallographic refinement packages restrain models to ideal geometry targets defined as single values that are independent of molecular conformation. However, ultrahigh-resolution X-ray models of proteins are not consistent with this concept of ideality and have been used to develop a library of ideal main-chain bond lengths and angles that are parameterized by the φ/ψ angle of the residue [Berkholz et al. (2009), Structure, 17, 1316–1325]. Here, it is first shown that the new conformation-dependent library does not suffer from poor agreement with ultrahigh-resolution structures, whereas current libraries have this problem. Using the TNT refinement package, it is then shown that protein structure refinement using this conformation-dependent library results in models that have much better agreement with library values of bond angles with little change in the R values. These tests support the value of revising refinement software to account for this new paradigm." - Donnie Berkholz from Bookmarklet
My latest paper just came out. The continued saga of improving atomic-resolution modeling of proteins. If any of you will be at ACA next week, come talk to me about it! - Donnie Berkholz
Congrat! Are you going to push this with Vriend (Nijmegen/NL) for their PDB reanalysis? - Egon Willighagen
Right now we're working on the main refinement programs — Phenix, Refmac, Shelx, TNT, etc. Hoping to get others to pick up remaining work once they're convinced of the value. - Donnie Berkholz
Congrats! - Ruchira S. Datta
Deepak Singh
AppCloud CLI | Engine Yard Cloud - http://www.engineyard.com/product...
Love it : "With the Engine Yard Textmate bundle, you can easily deploy to Engine Yard AppCloud via your Textmate environment." - Deepak Singh
Rich Apodaca
Deepak Singh
"drive forward an approach to computing that is fundamentally different from our previous approaches" - what would that look like? - Todd Hoff
Get rid of NFS once and for all - Deepak Singh from iPhone
And start thinking about extending a now well baked framework for our purposes - Deepak Singh from iPhone
Totally agree. But to be fair biology has a bit more complexity than social networks :-) - Adam Kraut
The real-time transactional nature of social networks helps event things up a bit :-) - Todd Hoff
Adam, agree there. All the more reason to jump in head first :) - Deepak Singh
We should always be taking the compute to the data. Actually we should be keeping the data always available for access by computing. And quite honestly, the kind of data that Yahoo is working on is not social network data, neither is the data that the guy from a large financial institution was running on a test cluster (1 PB), nor is all the geospatial and other data types that I see.... more... - Deepak Singh
Deepak Singh
The Biological Data Scientist - http://mndoci.com/2010...
I agree with one caveat - data science needs data scientists, but regular data analysis needs much easier to use pipelines. But there is a continuum here and we need all of these different skills sets. To pretend any one person can cover the needed spectrum is insanity. - Cameron Neylon
@Cameron... have you not read my blog post? Data analysis is not something you can do with 'easier to use pipelines'... data analysis is just complex. Theoretically, it is possible to visualize this complexity... but I'd say that's about the holy grail of data analysis anyway... that will surely get you a Nature paper. Model validation is an unsolved problem, and possibly cannot be... more... - Egon Willighagen
Egon, no I'm saying something quite different. There is a mass of data analysis that is very conventional and straightforward for which the tools are appalling and make the process much harder to learn. This is fitting of known models to known data kind of stuff. Absolutely agree with all of you where you are taking data, particularly new, large and complex datasets and trying to find... more... - Cameron Neylon
...or let me try that another way. That fact that it is hard, or even impossible to solve the general problem doesn't mean that we shouldn't make efforts to find good solutions and interfaces for specific cases where we can. - Cameron Neylon
@Egon - totally agree that 'data analysis is just complex', but as I spend about 95% of my time doing the finding, fetching, parsing, annotating, formatting and reformatting files, and trying to keep track of which files belong to whom, what question they're trying to answer, what other related data is available etc before I can get down to any data analysis, I'm all for... more... - Cass Johnston
@Cass... yes, that part certainly can get improved... I have done several studies in this field, and for example use Chemical Markup Language and Resource Description Framework to do these things... yes, the field even manages to get that go wrong :) That indeed is an area where workflow helps... workflows can help in other areas too... *but*, the point I was trying to make is, that... more... - Egon Willighagen
@Cameron: see my comment to Cass... it indeed is no black-and-white... some things can easily be automated... downloading SD files from PubChem... calculating QSAR descriptors can not. (Not because you technically cannot, but because the normalization of the input and filtering out the, umm, noise from the input requires manual intervention... or, putting this in a bioinformatics... more... - Egon Willighagen
And I am not saying this is the end of the story... I submitted a large grant application last year, and tuned this year, to address these issue. They *can* be solved. Just not with current approaches. Give me funding, and I'll show you how to do it :) - Egon Willighagen
Cameron, don't disagree with you. That's where things like Galaxy come in. But that is not data science as I define it. You're leveraging the data scientists work. The problem as I see it is that biologists, or the people who fund biologists, haven't figured out where the data science challenges lie and how much effort is required. One could go one step further and say that for... more... - Deepak Singh
Deepak has a good point, and I think it's part of Egon's issue too - you don't want biologists to become too limited by canned routines or pipelines. More and more all disciplines need a better understanding of how to work with data. However, there are levels to this, and the tools(and learning thereof) co-evolve with the data. So maybe canned routines would serve as something to... more... - Mr. Gunn
@Deepak: I think that part of the problem is that people think that if you throw Word at it, all the problems are solved. Too many scientists think that if you add a button for it, it will just all work out fine. The problem with canned solutions is, that the only work for canned input. And that's where hell breaks loose... there is no such thing as canned input in life sciences... - Egon Willighagen
This is an example 'canned routine' which will go horribly wrong in the long run: http://friendfeed.com/harijay... - Egon Willighagen
@Egon, indeed. But I think that for all such canned routines, no scientist should be completely trusting of them. if you don't put in some effort to understand the limitations, then you can't really blame the routine too much (assuming there is documntation). It goes back to my bug bear - computer tools are analogous to proper hardware (spectrometers etc) - if a scientists takes the effort to understand the use of a spectrometer, they should do the same for the computer program - Rajarshi Guha
"if a scientists takes the effort to understand the use of a spectrometer, they should do the same for the computer program" yeah, wouldn't that be something ... - Egon Willighagen
Agreed. The real challenge lies in building systems that let people work efficiently to the extent of their knowledge to understand what they're doing but not further. - Cameron Neylon
I should probably put this in context. I have spent most of the day writing a little GUI widget that will enable people to add a particular type of dataset together. Apparently no-one had thought it important to provide anything beyond a (partially documented) python command...I deal with a certain amount of "if you're not on the command line you're not a real scientist..." which gets tiring - Cameron Neylon
Protocols for data analysis can be followed or adapted depending on the researcher's skill level. It doesn't have to mean automation, because the user must still perform the steps themselves. A deeper understanding of the method can be gained through study and practice, especially if it is used often. This seems similar to wet lab protocols, which may be written by experts, and followed by novices even as they seek to understand it in more detail. - Mike Chelen
Cameron: Perhaps that indicates the great need for improvements in the quality of the protocols? If there were a standard method in a book or journal that could be pointed to, presumably it could have a far greater depth of documentation and references. - Mike Chelen
@Cass - Love your case, since I can relate to it ! @All - Data is information, and information itself must not make sense. Actually, do you remember the "doing science" discussion? Now, how does that fit into the picture of making sense of information, since some people (including article reviewers) might not consider this as "doing science". Having said that, I would love to blog about... more... - joergkurtwegner
+1 Agreeing with no caveats. I think academia should realize the potential of mentoring the next generation of students with understanding of the Data Science and its importance. A Data Science 101 should be a compulsory module in all biology / computer science / bioinformatics graduate course. - Khader Shameer
Andrew Perry
Deepak Singh
Inter-Datacenter Replication & Geo-Redundancy - http://perspectives.mvdirona.com/2010...
Deepak Singh
Pierre Lindenbaum
"Stack Exchange will now be free" http://blog.stackexchange.com/post...
Very interesting how structured they seem to feel they need to make it. - Cameron Neylon
Rajarshi Guha
Deepak Singh
The anti-powerpoint movement - http://flashingneurons.deepaks...
The anti-powerpoint movement
ROFL, like it, still, what else are people using ??? - joergkurtwegner
@joerg : I use inkscape - Pierre Lindenbaum
@Pierre I've recommended using Inkscape for designing conference posters to colleagues but instead they use, yes you've guessed it, PowerPoint. Totally the wrong job for the task, I mean who'd ever want to design A1 sized PowerPoint slides? - Dan Hagon
Does anyone use Prezi, or was that just a flash in the pan? - Matt Leifer
It's not the tool, it's the style. If I see another slide with 5-10 list items spelling out exactly what the person is saying... - mikepk
@mikepk Totally agree. If you've not seen it yet you should check out Garr Reynold's presentationzen.com particularly his Google talk - Dan Hagon
Link to Garr Reynold's Google talk: http://www.youtube.com/watch... - Walter Jessen
went to this talk by Venkat Subramaniam (author of Programming Scala) where he used multiple windows of Textmate and live code evaluation inside the editor. It was really so much more engaging than powerpoint - Hari
As mikepk says, it's the tool. I think the problem is with the standard cookie-cuttter presentations - Deepak Singh
and for the record, unless I have no option Keynote. In many cases, I'd rather use nothing or a whiteboard - Deepak Singh
We never use powerpoint in the lab- all lab meetings are chalk talks - Hari
I've done that ... some of my business meetings, esp with pharma, folks are often surprised when I ask them if I can use flip charts or a whiteboard - Deepak Singh
I use beamer too, which is great if you need a lot of math in your talk, but it is even worse than Powepoint for encouraging dull bullet-point riddled slides. - Matt Leifer
late to the party, but I mainly use Keynote. Fits my presentation style pretty well.. lots of pictures, analogies and examples instead of bullet-points.. - Daniel Jurczak
Also late to the party, but I'm a beamer user too. It certainly does create a lot of work for non-bullet point talks (I usually make up SVG or PNG images to embed in beamer slides) but it's very customisabe, good navigation and the way it works kind of makes you think very carefully about the structure of the talk which is no bad thing. - Ian Simpson from twhirl
Thanks for that link Walter. I think people radically underestimate the amount of time required for a good presentation. Time for planning, practice and refining (not just writing). - mikepk
Other ways to read this feed:Feed readerFacebook