73 replies to this topic

[maestro949]

The data is far less complete and far more inaccurate, I fear, than you might imagine.  How can a random person sitting in their bedroom or their mud hut curate data that they know little or nothing about?  If you want to properly curate drug screening data, a good start would be a mass spectrometer, which the average computer user probably lacks, not to mention the sample, if it still exists and hasn't further degraded.

Within a decade we'll likely have the 3D conformation of every protein that the genome can generate. The computational horsepower to traverse the myriad of interaction combinations between these is probably many decades away.

What I imagine is using human computation to tinker with these molecules allowing humans on their wireless $100 laptops to use their incredible visualization skills in order to find ways these proteins can interact with each other. We can build the physics toys / games for this and let the people do the rest. We could even provide financial incentive. Gives "a penny for your thoughts" new meaning, doesn't it.

[caston]

http://www.imminst.o...ST&f=175&t=5821

[maestro949]

http://www.imminst.o...ST&f=175&t=5821

My Response: Calorie Restriction

By modulating a single input parameter, longevity can be severely affected by 30-40%

What if we find a dozen or so more that stack with each other and CR and they each only contribute 5% each? That means there is a lot less damage accumulating. Toss in a few wins in cancer treatment, stem cells, heart disease and incremental improvements, layer in a healthy lifestyle (particularly excercise) and you have a significant potential increase in longevity. Once biomarkers show up in force we'll be able to detect the minute deltas in the metabolic process and identify their upstream transcription factors. Finding a winning combination that significantly delays aging and it's damaging affects is not out of the question.

It's highly plausible that metabolically adjusted mice will be winning the MMP in the future. The winning combo for us will likely not be one path forward but the aggregate of many small wins across all of the possible domains and options.

Immortalists should keep options open across the entire spectrum of possibilities. We need more time to be around to fix the entire system. It doesn't matter how we get that extra time.

[Brainbox]

Was my proposal to "grab the cow by its horns" to try something in a the context of a pilot to naive in your opinion?

[maestro949]

Was my proposal to "grab the cow by its horns" to try something in a the context of a pilot to naive in your opinion?

Not at all. It's exactly what I'm exploring. I am looking for frameworks from which to work in where we can start building out the model and layers even if it's what seems hopelessly simplistic at first :

.
.
.
With the leg bone connected to the knee bone,
and the knee bone connected to the thigh bone,
and the thigh bone connected to the hip bone.
.
.
.

With the systems approach we enumerate everything and simply start looking for the association points between each element in each list. The more associations we can make, the stronger the model becomes over time. It becomes hideously complex over time but that's where we bring in the big guns of machine learning, genetic algorithms, puzzle building algorithms, etc.

Would it be possible to select only certain aspects of the behaviour of cells in its environment by selecting only the interfaces that matter for the aging process to test / model these?

Absolutely. After enumerating the basic data elements across the various layers we focus on building outwards from the data elements implicated in the aging process and start linking them up. The causes of aging, the aging related metabolic changes, the genes involved, the damage that accumulates (e.g. amyloids, cancers, diseased tissues, inflammation types, etc). We stick all the data together and assign numerical values to them in various ways. We rank them, we put biological ages to them as to when they occur, etc.

While we build on all of this data we continuously look for ways to visualize it from different perspectives in order to find new and unique patterns that might give us some theoretical interventions or even predict what the missing pieces of the puzzle are.

Do we even have sufficient knowledge to make such a selection?

We have sufficient knowledge to start building the prototypes but the data is scattered all over the place and biologists have historically had a bad habit of describing data with much verbosity. We need it to be machine readable to extract any value out of it. I'd love to find ways to pull all of the known aging data together into a relational database and curate it for the next generation of upcoming gerontologists.

Would it be possible to start somewhere in the middle of the structural hierarchy?

It's probably the only place we can start. I propose that a good starting place is to enumerate all of the aging diseases, age related changes, causes of aging and damage that accumulates whether they be confirmed or theoretical. We could start mapping them all together as best we can and inch our way into more complex realms as time, data and our ability permits. Hopefully more experts might get involved from various fields to help overcome hurdles as they arise. There's always bribery and blackmail if they refuse

Like taking a simple generally representative organ, virtually remove it from its real environment and test / model all its (functional) interfaces? Thereby using a high level of abstraction as a reduction of complexity?
Or use both approaches combined?

Academic efforts have been trying this for quite awhile. The projects usually end when the grant money runs out or they hit the complexity wall. Rounding these up and compiling the data they gathered into a central systems model would be one of the goals of such an endeavor.

Edited by maestro949, 10 August 2007 - 06:13 PM.

[caston]

My Response:  Calorie Restriction



By modulating a single input parameter, longevity can be severely affected by 30-40%

That's true and it doesn't fight my analogy of comparing metabolism to data corruption either. You can build a computer with components that are far less suseptable to data corruption or operate the computer in an environment that is better for it in terms of heat, dust, electromagnetic interference, more gentle users. A CPU for example will surive longer if run at a low voltage but there is a limit to how low you can set the voltage.

What if we find a dozen or so more that stack with each other and CR and they each only contribute 5% each?  That means there is a lot less damage accumulating.  Toss in a few wins in cancer treatment, stem cells, heart disease and incremental improvements, layer in a healthy lifestyle (particularly exercise) and you have a significant potential increase in longevity.  Once biomarkers show up in force we'll be able to detect the minute deltas in the metabolic process and identify their upstream transcription factors.  Finding a winning combination that significantly delays aging and it's damaging affects is not out of the question.

Yes, definitely I once had to keep rebuilding a customers machine after it kept creating bad blocks on the HD. I changed nearly all the hardware (motherboard, cpu) and still every 3 months or so it would bad block the HD. I had been continuously RAing them. Eventually the problem was tracked down the optical drive and after replacing it with a DVD burner the problem never occurred again.

In evolutionary terms if this PC had to be self repairing and reproduce through a germlime it wouldn't have survived very long with a bung optical drive. But there could still be other components that create data corruption just at a much slower rate. If the rate of data corruption was so slow that it only made a difference to the the PC after it reproduced then it wouldn't be subject to natural selection. So there is definitely a lot of room for tweaking but you can only invest so much energy into finding an eliminating sources of data corruption. You will never be able to eliminate all of it.

Sometimes components added to help stop data corruption can actually cause it when it fails. For example a Hard Disk (HD) fan will keep a HD cool but it also could blow a lot of dust onto it and when the HD fan fails the heat will find it harder to escape and the drive will actually get hotter.

It's highly plausible that metabolically adjusted mice will be winning the MMP in the future.  The winning combo for us will likely not be one path forward but the aggregate of many small wins across all of the possible domains and options.

Immortalists should keep options open across the entire spectrum of possibilities.  We need more time to be around to fix the entire system.  It doesn't matter how we get that extra time

Yes, and remember in biology the software builds the hardware. If we backup the genetic code present in our stem cells and mitochondria we have a potential (system) restore point that could be used in the future. If we don't make that restore point we lose it forever as our hardware corrupts our data and we become infested with malware (viral inserts into ineucaryotic DNA) and system file corruption (cancer)

I propose we attack the problem using each approach as they are all valid to some degree in terms of technology, time and energy investment. If your going to do so much tweaking you need restore points.

Edited by caston, 10 August 2007 - 04:42 PM.

[niner]

Within a decade we'll likely have the 3D conformation of every protein that the genome can generate. The computational horsepower to traverse the myriad of interaction combinations between these is probably many decades away.

What I imagine is using human computation to tinker with these molecules allowing humans on their wireless $100 laptops to use their incredible visualization skills in order to find ways these proteins can interact with each other. We can build the physics toys / games for this and let the people do the rest. We could even provide financial incentive. Gives "a penny for your thoughts" new meaning, doesn't it.

Most of those proteins probably never see each other in the body, or if they do, their interactions are irrelevant. Having spent countless hours looking at 3D structures of proteins and other molecules on some extremely expensive graphics devices, I can tell you that it doesn't work this way. Protein-protein interactions are relatively uncommon. If you look for interactions between molecules that may or may not ever see each other in the body, you will generate a huge false positive rate, and the signal will be swamped by the noise.

Just because you have a gazillion entities doesn't mean you need to consider a gazillion squared interactions. Consider a flock of birds or a school of fish. Simply by understanding the interactions of one bird or fish with a few of its neighbors, you can generate the observed higher order motions of the flock. ("Emergent properties") You don't need to solve the N**2 problem.

It's highly plausible that metabolically adjusted mice will be winning the MMP in the future. The winning combo for us will likely not be one path forward but the aggregate of many small wins across all of the possible domains and options.

I think you are right about this.

[maestro949]

Most of those proteins probably never see each other in the body, or if they do, their interactions are irrelevant.  Having spent countless hours looking at 3D structures of proteins and other molecules on some extremely expensive graphics devices, I can tell you that it doesn't work this way.  Protein-protein interactions are relatively uncommon.  If you look for interactions between molecules that may or may not ever see each other in the body, you will generate a huge false positive rate, and the signal will be swamped by the noise.

Good points. I agree that we would want to try and constrain the searchable landscape to explore to a minimal set required. How about enzymes and proteins?


Edit: Fix typo, section deleted, moved to another thread

Edited by maestro949, 13 August 2007 - 11:42 AM.

[eternaltraveler]

wow a lot of new comments.

I'll get back to this once I get John his much needed report )

[apoptosos]

I find this topic amusing for two reasons:

1. This engineering approach sounds like it is fixing metabolism since the repair mechanisms proposed are a sort of synthetic metabolism whose purpose is to support or replace existing metabolic function. It seems that whilst the engineering approach is seeking to avoid an intimate understanding of underlying metabolism it is actually creating an alternative metabolism. The quest to identify and incorporate new enzymes that break down normally insoluble aggregates in mammalian tissues is an example.

2. It is fallacious to assume that there is less effort or knowledge-space that much be traversed in an engineering approach. The aspect of implementing an engineering approach can be just as resource expensive as "metabolic fixing".

All approaches are valid. The question is not whether one pursues the fixing of metabolism as opposed to an engineering approach, but rather one that has been asked by scientists and inventors since time immemorial: what is the least amount of data I need to obtain a solution.

[caston]

apoptosos: In terms of intervention what do you propose?

There is always of course the third option... that it may actually be easier to somehow transfer our consciousness to a new body that it is to fix our current ones...

[apoptosos]

Caston, I view all approaches as valid - despite their semantic variations - "engineering", "metabolism fixing", whatever their supporters choose to call them.. It would be a shame to discourage an aspiring student or researcher based on notions of quasi-scientific dogma. We must remember that great scientific strides were often achieved by aiming at targets that were not associated with the eventual trajectory of utility. If I were a gambler, I would place my bets on stem cell science, but that may have to do with my familiarity of this discipline. For a layperson to proclaim that one type of science is more right than another is not useful.

[caston]

moved my post to: http://www.imminst.o...=0

Edited by caston, 13 August 2007 - 12:44 PM.

[maestro949]

For a layperson to proclaim that one type of science is more right than another is not useful.

I fear we're all laypersons in this battle.

It'd be nice to find someone with "has cured aging in highly evolved species" on their resume however, no matter where you are on your learning curve in this field, everyone in front of you only has marginally more knowledge as it pertains to architecting a fix to even a small piece of the problem. Much of the experience needed to tackle aging still lies ahead of us whether it be failed experiments or efforts to try new and unique angles at coming at the problem.

We need more creative solutions than anything currently proposed. If we can't afford any of the current proposals, and it's unlikely that that money is going to be found, back to the drawing board. If the data's lacking and that looks impossible? Back to the drawing board.

We need more people studying the meme and tossing out ideas as to how we can get some small emergent ideas that are feasible from all perspectives (they don't rely on future tech, billions of dollars, etc) to bloom.