Theoretically you could have two DNA sequences that are substantially different yet code for the same result. I can think of a simple way and a much more difficult way to make this happen.
The simple way
There is a lot of redundancy in the genetic code. There are several different ways to code for the same amino acid sequence. So these two short DNA sequence code for the same amino acid sequence even though they differ at half of the base pairs:
The much harder way
There are a set of genes that code for t-RNA that determine how the the DNA codons will be read. If we could manipulate those genes then it would theoretically be possible to manipulate the genetic code in a synthetic organism. This would be akin to changing to a different computer programming language.
In such a synthetic organism you could have a gene for some sort of cell surface molecule with a sequence that is radically different from the sequence for the same molecule in a natural organism.
Sorry, I missed your post when I was replying above.
The problem with a programming comparison above the bare metal level is there is the element of compilation/optimization that occurs. But even if you wrote two programs in ASM that did nothing but populate some registers with values, discounting efficiency, there are plenty of ways to do that that will still lead to the same result. And as the scale of the program grows, so too would your flexibility.
But if you then took two of those first-generation crosses and crossed them, you’d get a mix: You could get some elephants that were completely normal genetically, but you could also get some in which both dwarfism traits were expressed, which might then be even smaller than either of the original populations.
And of course, which might die very quickly due to organ sizes being mismatched, or something like that. It’s a risky business.
Craig Venter and his teams are working hard to determine the minimal set of genes required for cell function. They have started by replicating an existing bacterial genome synthetically, and placing it into an empty cell, and restarting the cell to make it work and replicate.
Now, they are using the synthetic DNA process to identify which sections of the genome are required, and what they do. They hope to build a library of biological elements and a pattern of structure that allows them to create synthetic bacteria that can be used to do thing like create hydrocarbon fuels with high efficiency. But even with rapid DNA sequencing, there is a lot of testing involved, and it takes time for each sequence to be tested. Computers help analyse the data, but we are years, maybe decades, from the first fully artificial bacteria, let alone even a simple animal.
Other researchers have used bioengineering to create DNA that has synthetic elements - by repurposing redundant codons they have allowed cells to use new amino acids. This allows the production of novel proteins, and also means the cells rely on an amino acid that does not exist in nature - reducing the risk of such a cell escaping into the wild.
We know a bit about how DNA works, but there is a long way to go.
We can probably “engineer” an elephant that will experience retardation in growth, or an animal with three eyes, provided you don’t mind that one of the eyes is growing out of its hindsection and isn’t attached to the brain. Making the creature viable and the additions functional, however, is really, really, really difficult. In fact, it is so difficult that we cannot really claim to be doing “genetic engineering” at a fundamental level. With the exception of a handful of still-academic projects, all of the genetic modification we do is some form of chimarism; that is, identifying and splicing genes from one species into another to obtain some modification of expressed features, and even that is often a very trial and error process. Actually predicting how a gene will be expressed (which is largely controlled by factors that are actually external to the genome) much less synthesizing an entirely new feature from a tabula rosa is still substantially beyond the state of the art in genetic engineering.
By analogy, we are still in the Stone Age of genetic control and looking forward to reaching the equivalent of a rudamentary Bronze Age in, say, the next twenty or thirty years. We can take available resources and shape them to our needs in a very primitive fashion, but we can’t actually separate raw materials and produce useful tools. The basic building blocks of life are very simple, just as an alphabet is composed of a couple dozen basic symbols and variations thereof, but the grammar of life–which is defined by the interactions between proteins that are coded by genes–is so fantastically complex that even the fastest and most sophisticated computers can only simulate the interaction between the most simple proteins, and then only to a certain degree of approximation.
Not theory any longer. It’s been done, at least with one tRNA. IIRC, they actually made a hybrid version of the aminoacyl-tRNA synthase, which is the enzyme that attaches the amino acid to the tRNA. They took the “grab a certain amino acid” part from one gene and stuck it onto the “stick the amino acid onto a tRNA with THIS anticodon” part of another gene, and viola. You get amino acid X in the protein where you should get amino acid Y.
There’s a great book for the layman that covers this subject: The Art of Genes by Enrico Cohen (subtitled “How organisms make themselves”). It covers a good deal of what we know about how DNA works during the development of an organism, and just as interestingly, tells us how we learned a lot of it (including why so many human genes are named after diseases).
A great example is the noble fruitfly, from which we’ve learned so much. It shows how a lot of the organization of an organism is based on segments, differentiation between segments from head to tail, and then differentiation within segments (mostly, dorsal versus ventral) and how this differentiation is handled by just a few genes: genes very similar to human genes that have very similar roles during human embryonic development. (Which in itself is pretty astounding, given the “distance” between insects and humans.) These control genes lay down a pattern in the organism, guiding other genes and causing them to do rather different (yet similar) things in different places, such as making wings versus legs.
A fascinating book distilling the lessons of a number of highly specialized fields of study. Highly recommended. I read it when it was new, around 2000, and I’ll probably read it again before long.
Plus you get to learn why Goethe said that everything in a plant is a leaf, how he was laughed at by his contemporaries (“stick to poetry, bub”), but turned out to have made a very telling observation that wasn’t understood until the 1970’s.
As for “understanding the programming language”, well, a tremendous amount is understood. We may not understand what a specific protein created by a gene actually does or how it functions, but we can spot the protein code and the promoters for it in the genome pretty readily.
If there is a biological protein already out there we want to make, and we know the dna sequence for it, we can probably make it. The thing is, all those high level features you mention, such as fur or tails, are mediated by complex systems created by the DNA. It’s because the DNA is just a code to manufacture a computer system that will figure out how to actually complete a task.
You mention making acetaminophen using a spider. This would totally work…except that acetaminophen isn’t made by nature, and there’s no combination of amino acids that will make it, either. We’d have to design an entire chemical production chain out of proteins, if that is even possible for this molecule. We cannot yet design custom proteins to catalyze the production of something like acetaminophen, nor could we get all the switches right to make a cell produce it on demand like that.
However, if you wanted to make human growth hormone using a spider? That’s another story, and relatively straightforward.
So once we understand the programming language of DNA, how many code monkeys will we need to hire to write us our own Shakespeare? I estimate over a million for a piece of work like that man.
Not much. You need to add selection pressure. And the experiments have been done. One group has allowed millions of generations of bacteria to grow, with continual sampling. Their bacteria evolved the ability to digest citrate, which it didn’t have before. They were able to identify the key mutations which were not selective, but which allowed the final selective mutation to be sucessful. They could even rerun the process using frozen samples - once the key mutations were in place, the final mutation was basically inevitable within a few hundred generations. It gives some insight into the inevitability of things like antibiotic resistance, which is another experiment we have been running for a long time.
Random selection is slow, however. It does not give bioengineers all the tools they need, nor the selectivity they desire. Creating an appropriate selection pressure is hard and full of unintended consequences. So understanding the details of DNA engineering is far more important, to allow actual control and selection of all elements of a synthetic organism, even to ensuring it won’t survive in the wild.