AI
Friday, November 28, 2003 6:30:27 PM
Interesting discussion 
I wonder, though, if the analogy of program/data to coding/non-coding DNA is not stretched a bit thin as it pertains to finding disease genes. My understanding of programming is limited to html scripts and that's about it so I could be waaaay off, so bear with me.
I had always assumed that a bug in the software, however pesky, was fairly simple in nature (~Mendelian). A program could have a giant bug that is catastrophic or many little bugs that made it run poorly. It could take a long time to fix or a relatively short time. But the bugs, once identified, could be easily comprehended.
If we assume that the majority of genetically based maladies are not simple but complex, frequently occurring sequences then this magnifies not only the quantity of the data but the complexity and comprehensibility of the problem itself. I don't have much clue as to how many variables there are but if we go further and take into account other things like movement in the non-coding regions the complexity becomes massively exponential with each additional gene.
So IMHO, if I understand this theory correctly, its validity changes little in the search for disease genes as one wouldn't be able to reverse engineer the making of a disease just by following the coding mechanisms of one or even a few genes unless the disease was simple in nature. Thoughts?
Also, FWIW - I always assumed that DNAP's maps covered coding and non-coding regions. When they say their maps cover the whole genome, I interpret that literally.
I wonder, though, if the analogy of program/data to coding/non-coding DNA is not stretched a bit thin as it pertains to finding disease genes. My understanding of programming is limited to html scripts and that's about it so I could be waaaay off, so bear with me.
I had always assumed that a bug in the software, however pesky, was fairly simple in nature (~Mendelian). A program could have a giant bug that is catastrophic or many little bugs that made it run poorly. It could take a long time to fix or a relatively short time. But the bugs, once identified, could be easily comprehended.
If we assume that the majority of genetically based maladies are not simple but complex, frequently occurring sequences then this magnifies not only the quantity of the data but the complexity and comprehensibility of the problem itself. I don't have much clue as to how many variables there are but if we go further and take into account other things like movement in the non-coding regions the complexity becomes massively exponential with each additional gene.
So IMHO, if I understand this theory correctly, its validity changes little in the search for disease genes as one wouldn't be able to reverse engineer the making of a disease just by following the coding mechanisms of one or even a few genes unless the disease was simple in nature. Thoughts?
Also, FWIW - I always assumed that DNAP's maps covered coding and non-coding regions. When they say their maps cover the whole genome, I interpret that literally.
