ScienceSim

Intel has a new OpenSim grid called ScienceSim. Not a great deal there yet but looks good and it will be interesting to see how much LSL can be ported. I may try Second Inventory too. There are a few freebies (avatars, etc), some buildings, a planetarium (below) and a sandbox on the Einstein sim.

Virtual objects are not real: opportunity or threat?

I've been working on the markers rezzed when users touch a gene. The marker itself now has an extensive dialog that allows for customisation of size and colour of both the base and the particle stream. It is also possible to rez a simplified version of the generic sequence object, basically allowing you to take a record of that gene into inventory for future analysis.

I had to make the default marker larger so that it could be seen at a greater distance. The image above is a mockup with the marker surrounded by five proteins with which its gene product is believed to interact. These proteins, of course, are encoded by other genes, some closeby but others potentially some distance away. You will have to imagine the particle streams that show the interactions (which may, of course, be between pairs of the satellite proteins as well).

I started by making the protein spheres into sit hack teleports that conveyed the user to their gene's location on the genome. A somewhat disjointed and unsatisfactory experience.

I then decided to bring the remote parts of the genome to the site of the original marker by means of a map positioned above it. Not hard to do as I already have the textures, of course. I'm not sure how far I want to take this as a browser within a browser seems slightly perverse.

I'm reminded a little of Weinberg's Everything is Miscellaneous in which I believe he describes the consequences of physical objects being able to be in only one place at a time. This has ramifications, for example, for those who shelve and catalogue paper-based books. By contrast, digital media are miscellaneous and can be in as many places as you want, their metadata based largely on your favourite tags.

In Web3D it seems we have a halfway house. Objects can be replicated: each marker can have its own map and in principle its own set of mini-proteins (sculpties, for example) drawn from elsewhere on the genome. The relevant books in the library find you.

On the other hand, the genome layout is not arbitrary. Genes close together are more likely to share a common promoter and regulation and, indeed, to interact. Work in Caulobacter has shown the physical association of specific parts of the genome with the cytoskeleton as the replicating genome partitions between the dividing cells. I don't want to lose this pseudo-physical proximity effect.

Furthermore, one cannot neglect the impact of avatars. Having a small group looking at a shared map is one thing but 30 newbie avatars crowded round a teacher at the edge of a 40 m drop is distinctly non-ideal. I want to encourage them to spread out. As a side issue, making the markers unnecessarily complex is also bad design.

I'm tempted to say that the satellite proteins themselves should be arrayed on a miniature version of the chromosome but perhaps other ideas will occur in due course. Whizzing round on a tourbot of some kind is one possibility. Either way, the fact that objects represent physical entities but are virtual and hence readily replicated causes an interesting tension.

TB bug eats cellulose? nom nom nom

I'm starting to add a few proteins to the genome. The interactivity for all this still isn't worked out -- basically I'm playing. The first protein is encoded by the cel6/celA gene and is actually a functional cellulase, a rather surprising enzyme to be found in a human pathogen. The original paper speculates that this might be an evolutionary relic inherited from soil or plant-associated ancestors but comes down in favour of the notion that it either signifies a secondary host where cellulose is present or else the need to penetrate a cellulose-containing biofilm. Having this kind of conundrum right by the origin is a bonus.

At the moment there is a disk (red in the picture) to touch that rezzes the protein using the Rez-Faux API. After 30 seconds the protein automatically derezzes, the aim being to conserve prims. Eventually I might build this functionality into the gene markers.

Obviously some strategy (more likely strategies) for engaging avatars of various ilks is needed.

Web integration in SL: lost in translation?

In a fine blog post, Fleep Tuque chronicles the sometimes sad limitations of both first and second life. People have suggested that blogging may be in decline when faced with microblogging's opportunities for immediacy and community. Fleep's post, among many others from the SL community of late, reminds us why this should not be allowed to happen.

On a purely technical level, Fleep complains:

I can’t show a flash or .wmv movie in Second Life, can’t collaboratively access webpages and documents with others easily, and it takes forever and 50 steps to do something as simple as making a prim clickable to launch a webpage.

There is some truth in what she says but, on the recommendation of Carol and Ledoof, I have at last tried Damani's viewer (on left of image), not because it smoooths all the bumps in HTML-on-a-prim (HOAP) but in the expectation that it at least simplifies matters a little (as per Fleep's needs). On balance, I think it does.

One good point is that the viewer is trivial to setup and use. It is also commendably flexible, being able to read URLs from chat, bookmarks or a notecard. On the downside, it still can't follow links (the Daden and SLOODLE browsers work round this SL limitation) or scroll.

By way of compensation, it provides a direct link to chat-mediated Google search (shown in the image), integrated viewing of YouTube videos (no idea whether this uses native mp4 or translated mp3 files), listening to Shoutcast streams, and display of still images. Most importantly, the experience is touch-based and slick, at least once you realise the Damani logo at the bottom right returns you to the main menu! Thereafter it is simply a matter of using the arrow keys to navigate menus, entering search terms in chat when prompted.

I've addressed the issue of collaborative authoring previously and have nothing new to add (all previous posts tagged HOAP). However, the take-home message from the Damani product is that a smooth user experience is probably more important than heavyweight functionality. In that regard, a customised EtherPad would have much to commend it.

That said, overall I am still slightly perplexed by the Linden roadmap for web integration. The private sim-based and presumably heavyweight Immersive Workspaces version 2.0 product developed with Rivers Run Red hasn't clarified matters thus far. The collaborative authoring, I understand, is not inworld/immersive - I could be wrong but the demo sim (in the map above) is not accessible. Given the educational discount mentioned on the pre-Christmas Metanomics show, the absence of any substantive follow-up on the SLED list (tours, for example) is surprising.

While embedding the 3D browser in a web page is one option, I prefer the current inworld immersive experience with the opportunity to popup an integrated media browser when necessary (something Damani should add, by the way). However, while the integrated browser supports hyperlinks and scrolling, it doesn't get round the problems of Flash or Java other than via launch of a fully external browser.

Predictions as to the future seem to be largely a matter of speculation and one has to appreciate that web access and collaborative authoring are not priority issues for most gamers and SL residents apart from RL business and education.

It seemed at one stage that the integrated client would swap to WebKit with the work out-sourced in part to Nuanti. However, it is very unclear (to me, at least) how this is progressing and in the resulting vacuum the open source activity in the associated (and somewhat spammed) Google Groups seems to be leaning towards solutions based on the Google implementation of WebKit (Chromium) that underpins their new browser, Chrome. However, at the present time there is no support for Linux or the Mac which suggests additional delays.

As Fleep implies, this is a significant issue not only for individuals but also at enterprise level where I suspect IT directors can be expected to look with suspicion on any platform without a credible web interface.

Molecules with shadows

I previously bemoaned the absence of shadows to mitigate the snow-blinding effect of the decking. Well, I followed my own advice and downloaded the most recent Shadowcraft viewer from Kirsten's site and then followed the instructions on her blog and, remembering that Ctrl-Alt-D gets you the Advanced menu, this enabled me to grab a couple of screenshots (click through for full-size). As she says, the viewer needs high-end hardware and is a little unstable though I found it more responsive than I'd expected. While I would only use it for presentation images at present, it's one of many developments that vindicate open sourcing the SL viewer.

Adding small molecule structures to the genome

I filled in the centre of the Mycobacterium tuberculosis genome ring using ShapeGen so that I can build in the central area. Given that the centre has a 75 m radius, the effect is rather disorientating as you can no longer see the island below. Eventually, I intend dividing the area into quarters but as a temporary measure I have progressively shaded an outer ring so you have a vague idea where you are while in flight.

As a first step, I've added four anti-tubercular drugs, pyrazinamide, ethambutol, isoniazid and rifampicin (image above), adjacent to the genes that encode either their target (rpoB for rifampicin) or, in the case of pyrazinamide, the activating enzyme (the target is probably non-specific). The structures were obtained by chatting the smi code (from PubChem) to Hiro Sheridan's Orac structure rezzer, with the exception of the macrolide rifampicin which was abstracted from the pdb file 1icv using UCSF Chimera, converted to hin format using OpenBabel and then rezzed using Hiro's other notecard-based molecule rezzer. I still need to build the corresponding proteins as well as add further compounds.

From an aesthetic perspective, there is a distinct absence of shadow though there is development work going on in that area, I believe. As ever, there is a trade-off between hardware and features and some suggestion that the technology may not find itself into the official viewer any time soon. Fortunately, there is a third-party viewer with it called Shadowdraft which I hope to try at some stage.

If you want to see some more antimicrobials, do try the museum Hiro built on ACS Island. Hiro's rezzers are also available on the same sim.

The search facility has been marginally improved by having the marker rez in the general neighbourhood of the relevant gene rather than in the centre of the sector. I am, however, comprehensively failing to get to grips with SL rotations which I need to understand in order to determine locations more precisely. Ah well, there is a New Year's Resolution if ever I saw one.

The first genome tour

As you can see, I have set up a tandem that avatars can use to cycle round the genome. At a rate of 1.25 metres per second, it takes just under 2.5 minutes to do the complete circuit. I've started to write a commentary to go with this first, very general tour but it has the wrong focus at the moment, being science rather than tour-based.

I also got some way towards implementing a chat-based metadata search to complement the touch-based browsing. At the moment though, it only identifies which sector the hit is in and even then the particle stream is only visible from close up.

Finally, I've written a notecard that is available by touching the "Start here" sign. Lots more to do.