5 ms·

How many genes does it take to make a person?

46 points by Vigier 10y ago

spynxic 10y ago
Still seems like everyone is making attempts to understand genetics from a bottom-up approach. It doesn't make much sense to me to do so since there's no framework to talk about what's being produced by the DNA (i.e. protein functions) It's similar to giving someone a random series of numbers and asking them to continue it. The solution would be to try to understand the RNG instead of blindly pattern matching.
- jonlucc 10y ago
  Maybe I'm misunderstanding your frustration, but we can already produce many fully functional proteins (often in non-human cells) from their genes. It would be a magnificent feat to be able to know and produce an entire organism (familiar, not novel) from the letters on the page. Arguably, that's an important step to being able to do the kinds of designer genetics that people seem to be very interested in for making sure their kids are born without certain disease alleles.
- toufka 10y ago
  You are correct. There are two different 'componentry' arguments here hiding in the question. The first is essentially, how many separable ingredients does it require to make a person? This speaks nothing to how often, rarely, or combinatorially those ingredients are used. The answer seems to be a fairly consistent order of magnitude of ~10k unique protein components for a modern organism. With some retaining a lot of old or duplicated code that's been commented out, 'just in case'. The second (your point) seems even more valid. Many of those individual components are themselves just mix-'n'-matched of separable protein domains. And in fact the entire olfactory system generates thousands of separably functional proteins from a single 'gene' by mixing and matching compatibly functional protein domains. And on this count it's much more difficult to form an answer for the required components in a person. Especially since the DNA itself is not of much help when you're regexing for similar protein domains. If a production car contains 1,000,000 components to assemble, coming from 50,000 unique part numbers, with each designed part being built by extending/mixing-matching 2000 unique CAD files, using only 30 unique alloys/materials, how many 'components' does it take to make the car? The better question, exactly, is how many unique functions must the car accomplish in order to be a car. There is still the issue of scale in that, is 'being attached' as valid of a function as 'cooling the engine' or 'moving forward'. At least the counting of components can be done given today's understanding of genetics.
TheBiv 10y ago
This is one of those weird articles where having degrees in Biology, Genetics, and CS converges so bear with me as I explain a lot of domain specific terms Central dogma reminder: DNA -> RNA -> Protein. In the scheme of a person, a gene (DNA) doesn't mean as much as protein folding does. To give a programming example, DNA is the source code, RNA is the compiler and proteins are the executed program. A "person" is the executed program bc it's the end result. However tracing a lineage back to the source code of DNA to figure out the smallest amount of code required to create the program is often difficult (if not impossible). Genes are the blueprint, how those genes are transcribed into RNA and then translated into proteins matters FAR more than the amount of genes that it would take to make a sustainable human. So while I understand what this article is trying to get at, it's missing the central dogma of molecular biology which is still the wild Wild West!
- acomjean 10y ago
  I like the programming annolagy. What about alternative splicing where one gene can make multiple different proteins? The more I learn about biology the more clear it is how little we know.
- jonathansizz 10y ago
  You repeat a popular misinterpretation of the central dogma. The term was coined by Crick^, and actually says that once the information in nucleic acid goes into protein, you can't get it back out again. So if you want to diagram it, it would be DNA/RNA --> Protein, where the arrow represents information flow. Note that (rather ironically!) it was actually Watson who popularized the misinterpretation, in his college textbook. The reason I point this out is that there are many known violations of the Watson version of the central dogma, but none of the Crick version. Also, I'd take issue with your claim that genes are like a blueprint. I'd say they're more like a recipe, as stated in the article (and I think your comment implies this anyway). The larger point (that the article mostly ignores) centers around human vanity: we're actually not really much more complicated than many other species, especially our close relatives like the apes. Most of the differences between us and them are developmental, and you don't need many genes to produce such changes. ^ Crick 1956, 'Ideas on Protein Synthesis'. http://profiles.nlm.nih.gov/SC/B/B/F/T/_/scbbft.pdf http://profiles.nlm.nih.gov/SC/B/B/F/T/_/scbbft.pdf
nabla9 10y ago
>Extrapolating the analysis beyond the human knockouts study leads to an estimate that only 3,000 human genes are actually needed to build a healthy human. Genes are hyperparameters in the beginning of a dynamical process that creates a person. There is relatively straight forward path from DNA -> RNA -> primary structure of proteins. Complexity starts with more complex folding, secondary, tertiary and quaternary structure and their conformations and changes (can be affected by the state cell is in or environment) Another source of complexity is the gene regulatory network (GRN). GRN can be modeled as recurrent neural network or stochastic gene network. Genes are like neurons and their activity level is their state. Gene activity is modified by (noisy) activity or inactivity of other genes which have nonlinear inhibitory and excitatory effects on others. The state of the cell is dynamic recurrent process that can implement complex logic and respond to external and internal stimuli. How much it can learn and adapt is interesting question. Our microbiome, epigenetic factors, environment and the path person takes in life can change how genes express themselves trough this process.
coldcode 10y ago
There are more genes in corn than in people. But in this case the count doesn't matter, it's what you do with the genes.
roflmyeggo 10y ago
There is a simple counter-example that proves that this is the wrong question to ask. You cannot simply make a zygote with just the DNA from a sperm and an egg together alone. Both parents pronuclei are required for epigenetic reasons. If you try to do this, and the pronuclei are not present, then you will not make any viable offspring. This is just one example of how oversimplified the model taught by the book is, though it is a useful one. The only thing that is certain is that we are not anywhere near being able to claim to know how many genes it takes to make a person.
thatwebdude 10y ago
None, it's easier with them off (to make a person).

46 points by Vigier 10y ago

Still seems like everyone is making attempts to understand genetics from a bottom-up approach. It doesn't make much sense to me to do so since there's no framework to talk about what's being produced by the DNA (i.e. protein functions) It's similar to giving someone a random series of numbers and asking them to continue it. The solution would be to try to understand the RNG instead of blindly pattern matching.

jonlucc 10y ago
Maybe I'm misunderstanding your frustration, but we can already produce many fully functional proteins (often in non-human cells) from their genes. It would be a magnificent feat to be able to know and produce an entire organism (familiar, not novel) from the letters on the page. Arguably, that's an important step to being able to do the kinds of designer genetics that people seem to be very interested in for making sure their kids are born without certain disease alleles.
toufka 10y ago
You are correct. There are two different 'componentry' arguments here hiding in the question. The first is essentially, how many separable ingredients does it require to make a person? This speaks nothing to how often, rarely, or combinatorially those ingredients are used. The answer seems to be a fairly consistent order of magnitude of ~10k unique protein components for a modern organism. With some retaining a lot of old or duplicated code that's been commented out, 'just in case'. The second (your point) seems even more valid. Many of those individual components are themselves just mix-'n'-matched of separable protein domains. And in fact the entire olfactory system generates thousands of separably functional proteins from a single 'gene' by mixing and matching compatibly functional protein domains. And on this count it's much more difficult to form an answer for the required components in a person. Especially since the DNA itself is not of much help when you're regexing for similar protein domains. If a production car contains 1,000,000 components to assemble, coming from 50,000 unique part numbers, with each designed part being built by extending/mixing-matching 2000 unique CAD files, using only 30 unique alloys/materials, how many 'components' does it take to make the car? The better question, exactly, is how many unique functions must the car accomplish in order to be a car. There is still the issue of scale in that, is 'being attached' as valid of a function as 'cooling the engine' or 'moving forward'. At least the counting of components can be done given today's understanding of genetics.

TheBiv 10y ago

This is one of those weird articles where having degrees in Biology, Genetics, and CS converges so bear with me as I explain a lot of domain specific terms Central dogma reminder: DNA -> RNA -> Protein. In the scheme of a person, a gene (DNA) doesn't mean as much as protein folding does. To give a programming example, DNA is the source code, RNA is the compiler and proteins are the executed program. A "person" is the executed program bc it's the end result. However tracing a lineage back to the source code of DNA to figure out the smallest amount of code required to create the program is often difficult (if not impossible). Genes are the blueprint, how those genes are transcribed into RNA and then translated into proteins matters FAR more than the amount of genes that it would take to make a sustainable human. So while I understand what this article is trying to get at, it's missing the central dogma of molecular biology which is still the wild Wild West!

acomjean 10y ago
I like the programming annolagy. What about alternative splicing where one gene can make multiple different proteins? The more I learn about biology the more clear it is how little we know.
jonathansizz 10y ago
You repeat a popular misinterpretation of the central dogma. The term was coined by Crick^, and actually says that once the information in nucleic acid goes into protein, you can't get it back out again. So if you want to diagram it, it would be DNA/RNA --> Protein, where the arrow represents information flow. Note that (rather ironically!) it was actually Watson who popularized the misinterpretation, in his college textbook. The reason I point this out is that there are many known violations of the Watson version of the central dogma, but none of the Crick version. Also, I'd take issue with your claim that genes are like a blueprint. I'd say they're more like a recipe, as stated in the article (and I think your comment implies this anyway). The larger point (that the article mostly ignores) centers around human vanity: we're actually not really much more complicated than many other species, especially our close relatives like the apes. Most of the differences between us and them are developmental, and you don't need many genes to produce such changes. ^ Crick 1956, 'Ideas on Protein Synthesis'. http://profiles.nlm.nih.gov/SC/B/B/F/T/_/scbbft.pdf http://profiles.nlm.nih.gov/SC/B/B/F/T/_/scbbft.pdf

nabla9 10y ago

>Extrapolating the analysis beyond the human knockouts study leads to an estimate that only 3,000 human genes are actually needed to build a healthy human. Genes are hyperparameters in the beginning of a dynamical process that creates a person. There is relatively straight forward path from DNA -> RNA -> primary structure of proteins. Complexity starts with more complex folding, secondary, tertiary and quaternary structure and their conformations and changes (can be affected by the state cell is in or environment) Another source of complexity is the gene regulatory network (GRN). GRN can be modeled as recurrent neural network or stochastic gene network. Genes are like neurons and their activity level is their state. Gene activity is modified by (noisy) activity or inactivity of other genes which have nonlinear inhibitory and excitatory effects on others. The state of the cell is dynamic recurrent process that can implement complex logic and respond to external and internal stimuli. How much it can learn and adapt is interesting question. Our microbiome, epigenetic factors, environment and the path person takes in life can change how genes express themselves trough this process.

coldcode 10y ago

There are more genes in corn than in people. But in this case the count doesn't matter, it's what you do with the genes.

roflmyeggo 10y ago

There is a simple counter-example that proves that this is the wrong question to ask. You cannot simply make a zygote with just the DNA from a sperm and an egg together alone. Both parents pronuclei are required for epigenetic reasons. If you try to do this, and the pronuclei are not present, then you will not make any viable offspring. This is just one example of how oversimplified the model taught by the book is, though it is a useful one. The only thing that is certain is that we are not anywhere near being able to claim to know how many genes it takes to make a person.

thatwebdude 10y ago

None, it's easier with them off (to make a person).