At this stage we have multiple copies of each of the nested fragments of a given template and are able to determine which fragment ends with which base. All we need to do now is to order those fragments by length ,and then we will be able to recover the original sequence.
Recall that we are dealing with linear molecules at this point. Also,
DNA molecules when linearized have the property that they can move in
electric field at the speed approximately proportional to their
length. DNA is negatively charged in solution, so,
when we place it on a gel (they use acrylamide gel for sequencing
which has a very high degree of resolution) and place negative charge
at the one end and positive charge at the opposite end (see the
figure below), the fragments will start moving and the longer ones will go
slower at speeds approximately proportionally to their length. A laser
(for stimulating the emission of radiation) and detector (for
collecting the radiation stimulated) are placed a certain distance away from the initial
position. When a fragment passes the laser, it will
excite the fluorescent label which we attached to the terminator or
primer, and it will emit a signal at a certain wavelength depending
on the label, and the detector will detect it. Since we attached
different labels to each type of fragment, depending of the terminating
base, fragment ending, with, say C, will emit a signal at a different
wavelength than fragment ending with, say, A. Since we have multiple
copies of each fragment to start with, the signal will hopefully be
strong enough for the detector to record it, and, ideally, we should be
able to reconstruct the entire sequence based on the time order of
signal peaks.
Electrophoresis
The last figure is what the raw signal looks like directly
off the sequencing machine. This is a 4--component signal with each
component corresponding to the intensity of emission at a given
wavelength. The x-axis shows scan number or, equivalently, time.
Of course, in reality, we won't always be able to decode the
signal easily; or be
able to do this unambiguously. The signal may not be strong enough,
there could be contamination, the peaks may overlap, secondary
structures may form and change the relationship between time and length
etc. Hence we have the task known as basecalling.