While I agree that the bottleneck is elsewhere, I do not think they use the procedure you mention. If some of them do cycle sequencing, remember that it is not exponential (unlike PCR). There is no reason to use a thermostable polymerase for anything else here, and I would bet they do 'old-fashioned' sequencing with T7. And if they use restriction enzymes I would bet on SauIII because it might be used to generate more or less random fragments. Other might be useless, because they are not working with plasmids! The sequencing method you mention seem sth like Maxam-Gilbert, but you do not seem to remember it well. You do not need external standards (nor internal) for DNA sequencing, and they certainly use the Sanger dideoxy method (Maxam is usually used in footprinting experiments). Finally nobody does radiographies on film now. You just fluorescence detection in an automatic sequencing aparatus. I definitely agree with the conclusion, it takes more human-power than computer-power to complete the project.
I do not think such a huge computing power is going to be neccessary. Human brain-power will be far more important IMHO.
Depending on the sequencing methodology used, there are different approaches to assembling the sequence. As far as I remember the human genome has ben cloned into YACs, which may hold some 1,000,000 base pairs. If these are sequenced with a "shotgun like" method, they would generate some 20000 fragments, around 500 base pairs each. The whole sequence would be assembled by means of mathcnig 'overhangs'. If sufficient fragments are sequenced this should not be any problem at all, sth any desk computer could perform. Once all YACs are sequenced, they would be assembled into the 23 (+1?) chromosomes. This does not seem to be too difficult too. I see two big problems: 1. Debugging. If they use standard sequencing methods, the error rate may be as high as 0,1%. How are they going to cope with this? 2. Sequencing telomers or regions composed of repeats. This is going to be tricky.
My conclusion is: No distributed computing project is neccessary to accomplish the task.
Unfortunately I have never participated in an coordinated sequencing project, and all of the above are just my personal views.
Slashdot Mirror
While I agree that the bottleneck is elsewhere, I do not think they use the procedure you mention. If some of them do cycle sequencing, remember that it is not exponential (unlike PCR). There is no reason to use a thermostable polymerase for anything else here, and I would bet they do 'old-fashioned' sequencing with T7. And if they use restriction enzymes I would bet on SauIII because it might be used to generate more or less random fragments. Other might be useless, because they are not working with plasmids! The sequencing method you mention seem sth like Maxam-Gilbert, but you do not seem to remember it well. You do not need external standards (nor internal) for DNA sequencing, and they certainly use the Sanger dideoxy method (Maxam is usually used in footprinting experiments). Finally nobody does radiographies on film now. You just fluorescence detection in an automatic sequencing aparatus. I definitely agree with the conclusion, it takes more human-power than computer-power to complete the project.
I do not think such a huge computing power is going to be neccessary. Human brain-power will be far more important IMHO.
Depending on the sequencing methodology used, there are different approaches to assembling the sequence. As far as I remember the human genome has ben cloned into YACs, which may hold some 1,000,000 base pairs. If these are sequenced with a "shotgun like" method, they would generate some 20000 fragments, around 500 base pairs each. The whole sequence would be assembled by means of mathcnig 'overhangs'. If sufficient fragments are sequenced this should not be any problem at all, sth any desk computer could perform.
Once all YACs are sequenced, they would be assembled into the 23 (+1?) chromosomes. This does not seem to be too difficult too.
I see two big problems:
1. Debugging. If they use standard sequencing methods, the error rate may be as high as 0,1%. How are they going to cope with this?
2. Sequencing telomers or regions composed of repeats. This is going to be tricky.
My conclusion is: No distributed computing project is neccessary to accomplish the task.
Unfortunately I have never participated in an coordinated sequencing project, and all of the above are just my personal views.