Google Science Fair Project
Can computerised genetic algorithms help 'evolve' a cure for HIV?
See the full project:
https://sites.google.com/a/googlesciencefair.com/science-fair-2012-project-ahjzfnnjawvuy2vmywlyltiwmtjydwssb1byb2ply3qy4mg-da/home
My project utilizes advanced computer algorithms, molecular biology and the efficiency of evolution by natural selection to produce a practical solution to a pressing humanitarian issue. The ultimate goal of this project is to create a cure for HIV. HIV doesn’t kill, but rather renders the immune system incapable of functioning and thus allows secondary infections easy access the body. So by simply preventing HIV from infecting our immune system’s cells we can inhibit its effects.
The latest HIV research shows that certain mutations, in the genes of some individuals, essentially give them a higher resistance to HIV. However, this is only present in some racial groups and never completely effective. It occurred to me that if I used what we know about the DNA sequences of these mutations I could use an algorithm to assess how resistant any random DNA sequence would be to HIV. The algorithm could then use a simulated form of ‘natural selection’ to filter out the most effective DNA sequences from the numerous possibilities. Unfortunately, science does not yet know exactly what makes a DNA sequence feasible or lethal, but with the progress being made in human genome research it is very likely that we will soon have this information. For the purposes of this project, I fabricated a set of rules that I call a “Health Key” to dictate both the lethality and HIV resistance (health) of a DNA sequence. I also down-scaled the sample for my experiment to simplify and clearly illustrate the process.
With this in place, I had everything necessary for my algorithm; namely a “Health Key” and a set of parameters (to assess the effectiveness). I created a computer script using the Python programming language. The script generates random strings of 4000 to 5000 characters (digits 1-4) in length. This represents our ‘human genetic code’. The script tests the strings for lethality and healthiness and selects the best 50. The top 10 remain unchanged, while the remaining 40 are blended together (replicating breeding in nature) to create 90 offspring that are added to the remaining 10. The script repeats this process until a sufficiently ‘healthy’ string is found. I compared this to my original sequences (which had only one mutation) and in every instance the ‘HIV resistance’ was significantly higher.
Effectively, this could be scaled up to actual DNA sequences and a full human genome. If this was done, then we would possess a computerised DNA string that is immune to HIV. This same approach can even be applied to preventing numerous other gene based diseases too. Finally, by utilizing similar methods to those used by Craig Venter and his team in their synthetic cell project, we could then create immune system cells with high HIV resistance and, in doing so, severely limit global HIV infection.
See the full project:
https://sites.google.com/a/googlesciencefair.com/science-fair-2012-project-ahjzfnnjawvuy2vmywlyltiwmtjydwssb1byb2ply3qy4mg-da/home
My project utilizes advanced computer algorithms, molecular biology and the efficiency of evolution by natural selection to produce a practical solution to a pressing humanitarian issue. The ultimate goal of this project is to create a cure for HIV. HIV doesn’t kill, but rather renders the immune system incapable of functioning and thus allows secondary infections easy access the body. So by simply preventing HIV from infecting our immune system’s cells we can inhibit its effects.
The latest HIV research shows that certain mutations, in the genes of some individuals, essentially give them a higher resistance to HIV. However, this is only present in some racial groups and never completely effective. It occurred to me that if I used what we know about the DNA sequences of these mutations I could use an algorithm to assess how resistant any random DNA sequence would be to HIV. The algorithm could then use a simulated form of ‘natural selection’ to filter out the most effective DNA sequences from the numerous possibilities. Unfortunately, science does not yet know exactly what makes a DNA sequence feasible or lethal, but with the progress being made in human genome research it is very likely that we will soon have this information. For the purposes of this project, I fabricated a set of rules that I call a “Health Key” to dictate both the lethality and HIV resistance (health) of a DNA sequence. I also down-scaled the sample for my experiment to simplify and clearly illustrate the process.
With this in place, I had everything necessary for my algorithm; namely a “Health Key” and a set of parameters (to assess the effectiveness). I created a computer script using the Python programming language. The script generates random strings of 4000 to 5000 characters (digits 1-4) in length. This represents our ‘human genetic code’. The script tests the strings for lethality and healthiness and selects the best 50. The top 10 remain unchanged, while the remaining 40 are blended together (replicating breeding in nature) to create 90 offspring that are added to the remaining 10. The script repeats this process until a sufficiently ‘healthy’ string is found. I compared this to my original sequences (which had only one mutation) and in every instance the ‘HIV resistance’ was significantly higher.
Effectively, this could be scaled up to actual DNA sequences and a full human genome. If this was done, then we would possess a computerised DNA string that is immune to HIV. This same approach can even be applied to preventing numerous other gene based diseases too. Finally, by utilizing similar methods to those used by Craig Venter and his team in their synthetic cell project, we could then create immune system cells with high HIV resistance and, in doing so, severely limit global HIV infection.