I participated in a webcast of a public debate from the MIT Ray & Maria Stata Center this afternoon:

4:30-5:30pm
Public Debate
David Gelernter vs. Ray Kurzweil
Moderated by Rodney Brooks

Ray & Maria
Stata Center
rm 32-123, 32 Vassar Street
Massachusetts Institute of Technology
Cambridge, MA 02139

Are we limited to building super-intelligent robotic ‘zombies’ or will it be possible and desirable for us to build conscious, creative, volitional, perhaps even ‘spritual’
machines?

You can simulate a rain storm but no one gets wet. — David Gelernter

We can simulate biological neuron function. We can assume we will be able to simulate neuron function of 100 trillion connections at some point in the near future. Consciousness is an emergent phenomenon, but how does it work? At what point does consciousness emerge?

We think with our brains and our bodies, we’ll have to simulate both. — David Gelernter

The philosophical elements of my nature enjoyed the acknowledgment of the philosophical element required to define nature.

There were a considerable number of dropouts in the live stream, but an archived version is available on the web now. I saw a friend of mine in line to ask a question though the time ran out before he could get to the microphone. (Sorry Chris!)

BTW: what research group solved the protein folding problem, as Kurzweil claimed?

An interesting series of articles from the Economist: More special than you thought and Nature: Global variation in copy number in the human genome (PDF) describe a new appreciation for variation in the human genome.

From the Economist:

To find out how much things vary between people, they had previously focused on places in the string of molecular “letters” of which DNA is made where one letter is replaced by another in a significant fraction of the population. (These are known as single-nucleotide polymorphisms.) But it has been known for a long time that there is another sort of variation around. During the process by which DNA is copied, entire blocks of DNA can be accidentally deleted or multiplied. (Following such multiplied and deleted regions between the generations was an important genetic technique before the invention of cheap and rapid gene-sequencing technology.)

What the team has shown in the latest report is that duplication and deletion are much more widespread than was previously realised. Also, the duplication and deletion often involve active genes as well as the so-called non-coding part of the DNA, which is not translated into the protein molecules that keep cells running.

Does all that matter? The absence of a gene has obvious implications. Unless it is covered up by the presence of that gene on a sister chromosome (for chromosomes come in pairs; one from the mother and one from the father), trouble is likely to ensue. Multiple copies of a gene may bring more subtle problems. Genes pass their orders to the rest of the cell via messenger molecules copied from their DNA. Too many copies of a gene might mean too many messengers and thus too much protein. That might, in turn, cause disease.

The news is Copy Number Variation (CNV) is significant and more widely varied than previously thought. The HapMap database of 270 individuals from four populations representing Europe, Asia, and Africa, is only the beginning.

As an aside: we (CBRI) built a version of the HapMap data as a database. It took three days to load the phase I and phase II data. We hope our local researchers will find this a valuable resource.

It was in the low 20s this morning as I rounded Lake Harriet in South Minneapolis, MN. As I ran, the sweat from my body rose out of my fleece and formed a layer of frost on my shoulders.

The usual characters were there, the ducks, the dogs and their handlers, groups of ladies chatting, and old men walking alone.

We greeted each other with a knowing glance. We knew something exceptional had occurred. On a Monday morning, in the early light of day and new crisp bite of Fall. We were happy to share the experience, if only from our own orbits.

What a glorious run.

Lightweight data exploration in Excel seems to be a natural fit. But there has been surprisingly little visualization innovation in this tool since the beginning. A three-dimensional pie chart does not effectively convey any more information than a two-dimensional model. Most users of Excel in the Life Sciences appear to use it as a database or simple LIMS system.

There may be room in Excel to split out the data management and the visualization components. Until that happens, Chris Gemignani from Juice Analytics offers some interesting in-line bar charts, in the spirit of Edward Tufte’s Sparklines.

The follow-up article highlights some suggestions by readers of the original article. Two-columns used to show positive and negative values, gradient fills, dot-graphs, anchored dot graphs created by repeating dashes finished with the letter “o”, even in-cell Gantt charts.

For other gems like this, Juice Analytics is a site worth watching.

From the biodas.org website:

The distributed annotation system (DAS) is a client-server system in which a single client integrates information from multiple servers. It allows a single machine to gather up genome annotation information from multiple distant web sites, collate the information, and display it to the user in a single view. Little coordination is needed among the various information providers.

While I prefer the earliest versions, written primarily in PERL, by Lincoln Stein and Robin Dowell, there appear to be a number of DAS server projects from which to choose.

It has been suggested we look into the Dazzle server. Dazzle is implemented as a Java servlet, using the BioJava APIs. First pass reading shows an excellent pedigree as it was developed at the Wellcome Trust Sanger Institute by Thomas Down. A keen eye notices there has been no development on this project since the 31st August 2004.

Perhaps new development is waiting on the DAS2 specification? So far, I like what I see:

DAS 2.0 is implemented using a ReST architecture. Each document (also called an entity or object) has a name, which is a URL. Fetching the URL gets information about the document. The DAS-specific documents are all in XML. Other data types have existing widely used formats, and sometimes more than one for the same data. A DAS server may provide a distinct document for each of these formats, along with information about which formats are available.

DAS 2.0 addresses some shortcomings of the DAS 1.x protocol, including:

• Better support for hierachical structures (e.g. transcript + exons)
  
• Ontology-based feature annotations
  
• Allow multiple formats, including formats only appropriate for some feature types
  
• A lock-based editing protocol for curational clients
  
• An extensible namespacing system that allows annotations in non-genomic coordinates (e.g. uniprot protein coordinates or PDB structure coordinates)
  

The dissertation introducing the ReST architecture by Roy Thomas Fielding includes a rather interesting analysis of Java and JavaScript languages (section 6.5.4.3). I tend to agree.

People who manage Java-based production services deployed under Tomcat are wary of that environment. I’m looking for DAS deployment alternatives.

Currently the server of interest is ProServer DAS server also from the Sanger Center. ProServer is a lightweight DAS server written in Perl. It should be simple to install and configure and has adaptors for a wide variety of data sources. It claims to be easily extensible allowing adaptors to be written for other data sources.

ProServer DAS is under active development. Though not currently DAS2 compliant it will likely be one of the first to support the specification.