Biomedical Engineering Reference
In-Depth Information
case, the preparation of each target solution requires an extra dilution step.
The preparation plan requires 2
m
mixing-and-dispensing operations.
6.2.2 experimental results and Comparison
Next, we use the planning algorithm to carry out solution preparation for
protein crystallization.
For simplicity, we extract 24 target solutions from the thousands of solu-
tions for the experiment as listed in Table 6.1. From Table 6.1, we can see that
17 types of reagents are used, as listed in Table 6.2. After applying the solu-
tion-preparation planning algorithm, 17 source solutions with certain con-
centrations are chosen corresponding to the 17 types of reagents and stored
in on-chip reservoirs.
Next, we prepare these target solutions. First, a manual operation is used.
A pipette that can handle a minimum volume of 20 μL is used. Preparing the
target solution consumes 22 mL of reagent stock solutions and takes 1.5 h.
In contrast, the proposed chip design and the solution-preparation plan-
ning algorithm takes only 18 min and 12 μL of reagent solutions. For pro-
tein crystallization, reagent concentration is very important. Therefore, we
need to guarantee a high level of accuracy of concentration while preparing
the target solutions. For a digital microfluidic biochip, the key to generating
solutions with precise concentration is to maintain a constant volume of the
dispensed droplets.
Experiments have shown that our chip design achieves a high level of
consistency in the volume of dispensed droplets (variation < 0.5%), which
indicates high accuracy in the concentration of the prepared target solutions.
Note that the accuracy will degrade when multiple iterations of dilution are
carried out. However, results also show an error limit of less than 2.5% even
when five iterations of mixing-dispensing operations are used in preparing
the target solutions
(see Figure 6.19)
.
6.3 Chapter Summary and Conclusions
We have presented a multiwell-plate-based digital microfluidic biochip
design for protein crystallization. The proposed biochip is capable of con-
currently setting up 96 conditions, thereby achieving high throughput. We
have also applied an efficient algorithm to generate a pin-assignment plan
for the proposed design, which enables control of the biochip with only a
small number of pins. Compared to a directly addressable biochip, the pro-
posed pin-constrained design achieves a significant reduction in fabrication
cost. A testing and diagnosis technique for locating defects has also been
presented. We have also described efficient droplet-routing algorithms for
