tag:blogger.com,1999:blog-5506135718533366764.post8515202920286536507..comments2024-03-07T03:43:03.448-05:00Comments on RajLab: The hater’s guide to (experimental) reproducibilityARhttp://www.blogger.com/profile/13811773097412828786noreply@blogger.comBlogger15125tag:blogger.com,1999:blog-5506135718533366764.post-60597312534387363052017-06-30T13:09:11.527-04:002017-06-30T13:09:11.527-04:00We use mostly Google Docs. It works, not perfectly...We use mostly Google Docs. It works, not perfectly, but gets the job done. It has the advantage of being Google, meaning that it's unlikely to go out of business and disappear any time soon.ARhttps://www.blogger.com/profile/13811773097412828786noreply@blogger.comtag:blogger.com,1999:blog-5506135718533366764.post-46513855670375456622017-06-23T10:35:28.748-04:002017-06-23T10:35:28.748-04:00What online notebooks would you suggest?What online notebooks would you suggest?Anonymousnoreply@blogger.comtag:blogger.com,1999:blog-5506135718533366764.post-67379178910596611232017-04-13T08:15:01.298-04:002017-04-13T08:15:01.298-04:00You can do images, interactive 2d plots, even 3D v...You can do images, interactive 2d plots, even 3D volume visualization, you can interact with it, make it barf up a frozen pub ready paper whenever you want- the catch is you need to tell it what to do in code, but it's pretty easy to get a hold of the basics - here's an app somebody did for gels -- https://github.com/jfear/tapeAnalyst and here's a set of living demo notebooks, including a couple of biology focused ones (yeah, we geeks had a head start :-) ) https://github.com/jupyter/jupyter/wiki/A-gallery-of-interesting-Jupyter-Notebooks - we can talk more via email if you wantAnonymoushttps://www.blogger.com/profile/13956179722116046568noreply@blogger.comtag:blogger.com,1999:blog-5506135718533366764.post-73571880000214898142017-04-13T06:21:59.768-04:002017-04-13T06:21:59.768-04:00Agree 100% about the historical perspective. I'...Agree 100% about the historical perspective. I've brought this up a few times on Twitter, and people say, well, the future is this statistical perspective, but I've seen little evidence to suggest that so far, even though we've had big data and stats in biology for a while now.<br /><br />When it comes down to doing good quality science, as you say, there's just no substitute for hard, careful, critical thinking. No set of checklists can codify that. The best they can do is help out a bit with logistics.ARhttps://www.blogger.com/profile/13811773097412828786noreply@blogger.comtag:blogger.com,1999:blog-5506135718533366764.post-14813837640320904152017-04-13T06:18:14.858-04:002017-04-13T06:18:14.858-04:00Interesting, images for a notebook? Like gel image...Interesting, images for a notebook? Like gel images, etc.? We usually just stick those directly in a Google Doc. For large-scale microscopy images, we've just started using Dropbox Business with Smart Sync and have been pretty happy with it so far…ARhttps://www.blogger.com/profile/13811773097412828786noreply@blogger.comtag:blogger.com,1999:blog-5506135718533366764.post-45102565549220693572017-04-12T22:04:02.045-04:002017-04-12T22:04:02.045-04:00As another hater of the prevailing reproducibility...As another hater of the prevailing reproducibility discussion, I endorse all of your ideas for lab reproducibility. My first experience was in a yeast lab back before modern genomics - we had two computers in the lab, one to run Vector NTI, and another to write papers – and the good labs I knew were basically doing some version of your recommendations on paper, before the era of Google docs. Getting a cutting-edge experimental system up and running consistently is not trivial, and it requires serious effort & lab discipline to keep going. That also means it's not trivial for another lab that doesn't have the same system up and running to easily replicate a particular result.<br /><br />About intuition: One thing I find infuriating in overly broad discussions of reproducibility, dominated by psychologists and statisticians, is the lack of a historical perspective. Over the past 70 years, one of the most successful scientific fields - really one of the most successful in history - is molecular biology. And when you look back at how the field worked, people didn't rely much on statistical hypothesis testing, they didn't use large N in their experiments, and they did very little of what the Reproducibility crowd is arguing should be standard. And yet this has been one of the most historically successful fields. I have yet to hear anyone who claims that 'science' generally is facing a reproducibility crisis explain why molecular biology has been so successful without adopting the proposed remedies - or explain what has changed to make molecular biology supposedly less reproducible today.<br /><br />This is where I think your argument for intuition comes in. The pioneering molecular biologists had killer intuition, and they were relentlessly self-critical. That's how they made progress without adopting formulaic reproducibility standards.Anonymoushttps://www.blogger.com/profile/00474549004146223010noreply@blogger.comtag:blogger.com,1999:blog-5506135718533366764.post-20332083550706123182017-04-12T09:14:36.484-04:002017-04-12T09:14:36.484-04:00You're right, I should just give it a go. Than...You're right, I should just give it a go. Thanks for the pointer/plug though.Darach Millernoreply@blogger.comtag:blogger.com,1999:blog-5506135718533366764.post-52933102488127496022017-04-12T08:32:14.852-04:002017-04-12T08:32:14.852-04:00Well, we're about to go way off topic on this ...Well, we're about to go way off topic on this thread - I can respond here or give you an email, Arjun Raj can make the call - in short, its a browser based solution, its excellent not only at pictures but also whizzy things like animated charts and even volume visualizations :-) In the interim, check the link I gave to Jupyter, it has a very active community<br />Anonymoushttps://www.blogger.com/profile/13956179722116046568noreply@blogger.comtag:blogger.com,1999:blog-5506135718533366764.post-6015739067273385762017-04-12T07:56:54.696-04:002017-04-12T07:56:54.696-04:00I'm curious how it handles images, I'm ass...I'm curious how it handles images, I'm assuming that's trivial to load and print one, but can it store them standdalone in the output? So that it can be distributed independent of the filesystem?<br /><br />How do you keep these stored, synced, and shared with people?Darach Millernoreply@blogger.comtag:blogger.com,1999:blog-5506135718533366764.post-66025236634705197662017-04-11T21:23:59.396-04:002017-04-11T21:23:59.396-04:00:-( But paper is not computationally fungible :...:-( But paper is not computationally fungible :-( Anonymoushttps://www.blogger.com/profile/13956179722116046568noreply@blogger.comtag:blogger.com,1999:blog-5506135718533366764.post-24578040930033894692017-04-11T20:04:16.555-04:002017-04-11T20:04:16.555-04:00Well, for a while, people were using Google Docs i...Well, for a while, people were using Google Docs in the lab, but frankly, I think most of the lab has moved back to paper, sadly. I think Google Docs just ended up being a bit too cumbersome in the end… though I miss them dearly as a PI.ARhttps://www.blogger.com/profile/13811773097412828786noreply@blogger.comtag:blogger.com,1999:blog-5506135718533366764.post-20774253403002223192017-04-11T10:52:36.472-04:002017-04-11T10:52:36.472-04:00My 2c - Jupyter - cell oriented notebook in brow...My 2c - Jupyter - cell oriented notebook in browser, hosts a wide variety of back ends - my bestiary includes python 2,3, R, SBCL, and some even odder back ends - go here -- http://jupyter.org/Anonymoushttps://www.blogger.com/profile/13956179722116046568noreply@blogger.comtag:blogger.com,1999:blog-5506135718533366764.post-14285806861126396192017-04-11T10:31:39.852-04:002017-04-11T10:31:39.852-04:00For point 1), what do y'all use for online not...For point 1), what do y'all use for online notebook? Been doing google docs, but unhappy with loading / download-PDF times.Darach Millernoreply@blogger.comtag:blogger.com,1999:blog-5506135718533366764.post-38792140644644921582017-04-08T15:46:17.739-04:002017-04-08T15:46:17.739-04:00Haha, yes, not that we do all of this stuff either...Haha, yes, not that we do all of this stuff either! Agree that every step in the right direction is a good one, no matter how small.ARhttps://www.blogger.com/profile/13811773097412828786noreply@blogger.comtag:blogger.com,1999:blog-5506135718533366764.post-11674456785456058292017-04-08T15:39:11.439-04:002017-04-08T15:39:11.439-04:00As a computing guy in and out of machine learning,...As a computing guy in and out of machine learning, and most likely earning the derision of biologists everywhere for nursery school level oversimplifications...... don't admire us computational types too much - far too often that's what we say, not what we do :-) That said, I agree wholeheartedly with what appears to be your goal, from a non-biologists perspectiveAnonymoushttps://www.blogger.com/profile/13956179722116046568noreply@blogger.com