It sounds like the major problem with this technology is controlling the rate of passage of a single DNA strand through the detection pore. Instead of trying to solve that "hard" problem, why not design the system so that you don't need such tight control over the speed of the DNA strand?
In the current system, if the strand moves to slowly between reads, a base will be scanned twice. If the strand moves too quickly, some bases may be skipped altogether. You could slow down the rate of strand passage relative to the scanning frequency, but then you couldn't differentiate between a sequence of 3 G's in a row or a single G getting scanned three times.
If you design your DNA reader with multiple reading points in series (i.e. read the strand simultaneously at multiple points along it's length), this problem would go away. Here's how it works:
1. You assume that the entire strand moves at the same rate (this rate can vary, but must be slow compared to the scan rate of the base readers).
2. If any of your serial DNA readers record a change in base, you interpret this to mean that the DNA molecule has moved down the pore by one base height. Any detectors that did not record a base change are likely reading their next base as well, but it just happens to be the same kind as the one they read just prior.
By allowing for multiple scans per base, you increase the likelihood of making a correct call. In fact, you may be able to distinguish a C from a methyl-C from a hydroxymethyl-C and get epigenetic information at the same time you get sequence information! By using multiple detectors in series, you are able to detect when the DNA strand has moved one base height and get another crack at making sure you read the right bases.
If anyone from IBM sees this and thinks it might work - drop me a note. I would be very interested in participating in the development/testing of this technology.
- Cancer genetics researcher / clinician.