Comment by jacquesm | Hacker News Reader

Comment by jacquesm | Hacker News Reader

Parent Full thread

If only it were that simple. We're very far away from understanding how a particular DNA string relates to phenotypes. All we've found to date is a bunch of configuration bits and maybe a general idea of how things might be related to each other but not much more than that.

The gap in understanding between 'here are a bunch of bits that appear to do 'x'' and 'x' is vast, far larger than the gap between say being able to build a compiler to have an executable perform in a certain way.

If anywhere that is where the real breakthrough will be, in being able to effectively map the chain of binary bits in a genome to a certain phenotype in a mature organism without any 'leftover bits'.

Once you have that chain of causality mapped out in enough detail to clearly link all causes with their effects you can start engineering something to act in a way that is far outside what we normally find in nature.

I'm not sure if I'm expressing myself clearly about all this, maybe someone more versed in the matter can chime in but the path between DNA and organism is a lot longer and less well understood than the path between a low level language program and the resulting output of that program. Just having a 'compiler' will not change much about that understanding, though once we have that compiler we may be able to advance our understanding by a more direct approach in modifying pieces of an existing genome to observe the effects in the organism. For sure it will speed things up.

edit: if we did have that understanding it would be trivial to create a digital representation of a (simple) organism from it's sequenced DNA. The fact that we are nowhere near capable of doing this is a simple proof that merely being able to create a certain sequence of bits will not give us the capability to create organisms to order.

Depends on what type of organisms are we talking about. We do have enough understanding to target bacteria with a "compiler" approach. That's the gist of synthetic biology and the stuff that's being done around iGEM competition.

Bacteria are "simple" enough that we can just append some DNA sequences and have them develop stuff we want. It's predictable enough that people already created "standard library" of DNA blocks - so people can work with bacteria in a kind of dataflow environment, with underlying DNA representation hidden. You connect a "molecule sensor" block with an "amplifier" block, an "inverter" block and a "light up" block, and you get bacteria that start to glow in presence of some metal, etc.

But of course, once we move to the level of multicellular organisms, the complexity explodes. Here, as you say, we basically know that something in an unaltered system sorta correlates with a particular phenotype and if we change it then maybe it won't blow up.

Sorry, yes, I was assuming 'multi-cellular organisms' as the context because of the bamboo rake comment at the beginning, but you're absolutely right that for single cells the situation will be different. Even so, if we're going to expand the scope downward we might as well go all the way, viruses, arguably even more simple than single celled organisms are still full of surprises.

Sorry, yeah, I work with bacterial genomes and pretty much skipped over the bamboo comment.

>Just having a 'compiler' will not change much about that understanding

The compiler is called "genetic engineering" and it's a 30 years old field. In particular genetic engineering allows you to knockout specific genes and observe the phenotype. This technique is commonly used on bacteria, plants and animals. There is an ongoing mouse knockout project, an attempt to study every mouse gene by turning it off. It is 50% complete so far. By the way, the experiments are performed mostly by hand.

>We're very far away from understanding how a particular DNA string relates to phenotypes.

This problem can be solved much faster with a large scale effort. Thousands of automated experiments can run in parallel while recording the phenotype of the animals (including results of behavioral tests) and storing it for later analysis. On this genotype->phenotype dataset one can train an ML model. Actually there is a company that is trying this approach (just ML part, without automated experiments) http://www.deepgenomics.com/

I doubt that one company will be enough, though. With serious funding there could be much more progress in this area. We could have a full knockout map of mouse genome in less than a decade if we really wanted.

Reply on news.ycombinator.com