Disqus - Latest Comments for cariaso

Re: Pesky ADRs

cariaso — Sun, 18 Mar 2007 17:47:50 -0000

Patients that demonstrate the ADRs should be extremely interesting to the pharma/FDA. Has anyone began to track and resquence relevant segments of these patients genomes, to determine what genetic markers may explain or predict? Or at least biobanking their samples for a suitable day?

Re: The Proteins Wiki

cariaso — Fri, 22 Jun 2007 22:51:25 -0000

Auto annotating semi-structured data is not particularly hard. It can be done with any 'bot', such as those written with pywikipedia or any of several perl interfaces.

I maintain SNPedia a wiki based on mediawiki, but not part of Wikia, which uses a mix of bots and human effort to catalog the effects of human genetic variations. I've used a bot to scoop parts of GeneRIF and other resources into the system. I've used the same techniques for several other non-public wikis as well.

The relevant code is remarkably simple. If anyone has similar needs I would be happy to provide code or tips.

Re: EC2 as a scientific tool

cariaso — Fri, 31 Aug 2007 13:18:03 -0000

RunBlast.com is my effort in this space. I'm very curious to hear what any others may be doing inside ec2.

Re: Collective Intelligence in the life sciences

cariaso — Fri, 12 Oct 2007 19:24:55 -0000

Perhaps you will be pleasantly unsurprised to learn that the author Toby Segaran has been the Director of Software Development at Genstruct an AI based biotech.

Re: If you haven’t seen xkcd you must

cariaso — Sun, 09 Dec 2007 23:58:42 -0000

I'm declare my xkcd fanboydom
http://cariaso.livejournal.com/33797.html

Re: The web as platform: Amazon continues its march

cariaso — Wed, 19 Dec 2007 21:44:06 -0000

I won't even try and be subtle.

this is what RunBlast.com is all about.

http://www.RunBlast.com/wiki

Re: Blast - The Amazon version

cariaso — Wed, 19 Dec 2007 23:25:35 -0000

It most certainly is. And if you want to see part 3 of the series it will be delivered at PyCon 2008 Chicago by

Peter Skomoroch
http://us.pycon.org/2008/profile/0773dc/

for his talk
http://us.pycon.org/2008/conference/talks/?filt...

131. MPI Cluster Programming with Python and Amazon EC2

Amazon EC2 may offer the possibility of high performance computing to programmers on a budget. Instead of building and maintaining a permanent Beowulf cluster, we can launch a cluster on-demand using Python and EC2. This talk will cover the basics involved in getting your own cluster running using Python, demonstrate how to run some large parallel computations using Python MPI wrappers, and show some initial results on cluster performance.

RunBlast is going to be available as a download soon, and will allow you to run Blast and other apps on a virtual EC2 cluster. Oh, by the way, RunBlast uses the Amazon payment system. Wait, isn’t Blast free? Why should I pay for it? Well if you run Blast once a year, perhaps you shouldn’t be using RunBlast, but if you’re a regular user, and do some serious crunchching, this beats buying a cluster anyday.

People who mainly use the NCBI service will do well to continue doing so. This is really targeted at groups who have a small cluster inhouse which is mainly used to run blast.

When your needs have outgrown the NCBI webserver, RunBlast allows you to have that cluster without the big upfront purchase, or the hassles of maintaining latest versions of major tools and databases.

I actually use it to run bioperl, emboss, and MrBayes. NAMD has been on my todo list but hearing its of interest should bump it up. I'll drop a line when its available.

Re: Of tagspaces and finding information

cariaso — Fri, 21 Dec 2007 04:52:54 -0000

what chars are legal in a hash tag?
by excluding '=' you can support something extra

Tags are a sort of implied boolean
#sunset #beach
is equivalent to
#sunset=True #beach=True

but this leaves open the possibility of tags such as
#sunset #beach #temp=84F #camera='cannon powershot'

Re: MPIHMMER

cariaso — Sun, 06 Jan 2008 16:41:52 -0000

Well I haven't been able to solve it yet, but my notes on the process are at
http://www.runblast.com/wiki/index.php?title=Hmmer

and perhaps they will be useful to you or someone.

Re: Around the web - March 22, 2008

cariaso — Sun, 23 Mar 2008 11:09:52 -0000

Given your interest in semantic web, you may be interested to know that SNPedia joined the semantic web this weekend.

This is visible from every snp page, such as
http://www.snpedia.com/index.php?title=Rs4420638

The new footers shows it now understand categories a bit better, and this allows me to make queries such as

http://rurl.org/lmu

deeper integration into SNPedia and with sites such as uniprot and the ncbi will happen in time.

Re: Around the web - March 22, 2008

cariaso — Sun, 23 Mar 2008 11:11:22 -0000

and I meant to add, I hope to meet you at bioitworld this year.

Re: The Big Switch

cariaso — Tue, 10 Jun 2008 20:04:04 -0000

this is like chocolate and peanut butter. congrats to Deepak and congrats to AWS.

Re: Your personal health: 23andme v2

cariaso — Tue, 09 Sep 2008 02:58:26 -0000

"Our scientists spent many months sifting through scientific journals and online databases to select new SNPs". 23andMe's original chip was an Illumina550 plus additional custom content. This chip may be based on the 1M chip, but if it were only that they would have lost some of their existing advantage over deCODEme. I won't be surprised to find that all of SNPedia (from before a certain date) is part of the new content.

Re: Just Downloaded a File Containing my Raw Genome Data. Now What?

cariaso — Tue, 20 May 2008 00:46:16 -0000

I've registered and uploaded a genetic profile. It wasn't clear until after the upload, but I now see the message:

"Hi ,
Upload your data here and in a few weeks you can analyse your data usingMyFamilyHealth tools and services for FREE! "

Since the upload also had a text box labeled:

"Please tell us some detail of your data, such as:
i. From where did you acquire this data.
ii. The data format or specification.
iii. Any related details. "

It would seem likely a human (mayhap swswsu) will read that file and do your analysis.

Re: Just Downloaded a File Containing my Raw Genome Data. Now What?

cariaso — Wed, 21 May 2008 09:08:12 -0000

I've gotten a personal email indicating that they are working on it. I'll share what I learn.

Re: Just Downloaded a File Containing my Raw Genome Data. Now What?

cariaso — Thu, 29 May 2008 01:40:00 -0000

I have gotten the results back. The first thing I should mention is that while I greatly appreciate people who are so open with their genome, I didn't actually upload my own data, I uploaded data from 23andMe's Greg Mendel profile. The company behind this effort is Genseq which is based in Malaysia. There appears to be 24 public user profiles on the site, Although a few of those appear to be dupicates or test users. I'll be curious to hear if anyone can see the 'cariaso' profile, since I've done nothing to share it. A lot of the site emphasis seems to be on creating family trees linked to a manually entered medical history.

The UI is a little rough, but functional. Some of the artwork is quite good, but the look&feel of the dynamic elements isn't nearly as pretty. There are several visualizations, a relative risk which does not seem to render for me, a sized/colored ATCG (which seems a poor choice for what I think it wants to convey) as well as a 'visible human' style map which responds to mouse-overs and clicks. The last element is particularly well done, but seems more eye candy than scientific rigor. Notably visualizations in one of the main windows disappears when elements are moved outside the bounding box.

They appear to track 22 medical conditions which are classified as either high or low confidence, and for each there is a red/yellow/green high/med/low risk assignment. These also provide a % risk for european/asian/african populations vs 'my' data. Generally the site does link to pubmed, to backup the claims. I find the lack of an rs# makes it difficult to verify specific claims, but that is partly my bias showing through.

It all seems reasonable enough, but I'm not exactly overwhelmed. In fairness I've got a certain bias due to promethease, which is often similarly criticized.

Re: GINA Series: When Employer Genetic Testing is Appropriate [Page 3]

cariaso — Wed, 28 May 2008 00:57:55 -0000

GINA seems to prevent employers from testing for
http://www.snpedia.com/index.php?title=Rs3892097
to identify the 20% of employees who are at 3x+ risk of Parkinson's if they are frequently exposed to pesticides.