Hi Bill, I think I understand what you were saying. I believe you misunderstood the point I was making in my discussion with Meyer. I was saying, and Meyer agreed, that cells do not necessarily look like machines because they use processes that are foreign to machines. An example - when you shut your computer down, you just push a button. Cells rip out the button and also the circuitry. This isn’t bad design, it’s no design, at least if “it looks like a machine” is the formula.
Cells are replete with stuff that mocks the machine analogy.
I agree that they are certainly different in their mechanical operation but a cellular regulation system is conceptually the same as a human designed system. We are observing a set of parts that perform a specific function like cell division.
If I define scratch as a random sequence then there is no way for you to know this. I accept that the sequence was a non coding sequence that was altered to become a functioning enzyme.
What alterations to the non coding sequence were required?
I don’t know how many were required, but at least 7 happened. The link to the essay in the new Pandas Thumb is here. In case anyone wants to review it.
Not in my experience. Biological systems are very interconnected and messy. Many ‘parts’ perform different functions at different places and various stages. There’s also tremendous cross-talk between systems – For example, many proteins have multiple phosphorylation states and these can be manipulated by a host of other enzymes. Some of the phosphorylation states seem relevant but often much of it seems like noise or the result of promiscuous, non-specific activity. This hard-to-disentangle nature is what makes cancer research particularly hard. Same for immune-system disorders and more. If more things had specific functions, like we find in human engineered systems, it might be easier to address cancer and other medical disorders.
How is a telephone number sequence a valid model for protein sequence? That’s the real question. Do the digits in a telephone number interact with each other to create physical tertiary structures? If not, I don’t see how they are comparable.
To get to a tertiary structure you need a sequence capable of that fold. Sequences capable of that fold exist is sequence space the same as numbers that can call anyone in the world exist in numbers sequence space.
So how do the numbers compare between functional protein sequence and telephone numbers? Also, you need to add in biological reproduction where each generation starts with functioning proteins. How do you model changes to functioning proteins with telephone numbers?
You compare the amount of function that exists inside that sequence as compared to the total search space. This is what Art’s discussion was about in his 2003 article.
In the case of phone numbers it depends how you define function. Calling a specific person at one end and calling anyone in the world at the other.
At the end of the day with long sequences the only known cause to generate such a sequence is a conscious mind. This is the design argument.
Trying to include random mutation and natural selection is a viable cause has been difficult.
So how do you determine if any protein sequence has function? That seems like an almost impossible task given the millions of possible substrates a protein can bind to or alter.
If I randomly typed in 1 and then 10 digits into my phone, what are the chances that I would connect to another phone? I don’t think it would take too many tries before I found a phone number that worked. Also, if I start with an already functioning phone number the chances probably grow from there since I would have a fuctioning area code at a minimum.
This is indeed true. In the most basic case you can measure if a protein has a sequence that can fold as this is a minimum requirement for function. In the case of the Szostak experiment which was part of Hunt’s paper (I think) he measured the probability of a 70 AA protein binding to ATP which is a requirement of some proteins. Again, you cannot build a protein complex on “any function” as all biology is built on interdependent components.
You are exactly right. 30 trials assuming 300 million 10 digit phone numbers. The problem is when you stretch the sequence. If you go to 13 digits because you are making the call from France then you need 30000 trials. You can now see how the problem gets exponentially worse with sequence length.
Even then, there are many examples of functional proteins that do not have traditional folds.
I’m not following you on this one. On first glance, it would seem that a longer random sequence would have a much better chance of having a sequence within it capable of producing a fold.
I plugged this into SWISS-MODEL to see if there were any possible protein folds. The algorithm predicts at least 2 alpha helical folds in this random sequence.
That would seem to be a reasonable prediction for evolved proteins, yes. I based my statement simply on the basis that a protein with multiple functions can be far more efficient than independent functions. Granted, I’m a statistician not a biologist, but I’ve tinkered with genetic search algorithms, and I do not see how it would be possible to stop such interdependencies from forming if it favors fitness.
I’m with Dr. Swamidass on this. A sequence does not need to be optimal, it only needs to be functional. Sequences that can evolve must have some inherent unreliability as a source of variation.
Depending on where we draw the line for “life”, I wouldn’t necessarily require cell division, only a sustainable process that reproduces itself.