Hardware Reference
In-Depth Information
Due to the complex and randomness component interactions that are ubiquitous in
biological/chemical processes, predictive modeling and accuracy control become
difficult [
5
,
6
].
In addition to manufacturing defects and imperfections, various faults may also
arise during bioassay execution. For example, DNA fouling may lead to malfunction
of multiple electrodes in the biochip and excessive actuation voltage applied to
an electrode may lead to breakdown of electrodes and charge trapping [
7
-
9
].
These faults are hard to detect a priori, but they occur during bioassays in certain
situations [
9
]. Yet, despite such inherent variability, many biomedical experiments,
such as clinical diagnostics and drug development, require fluid-handling operations
that are highly accurate and precise. Each step in the protocol of a biochemical
experiment has an “acceptance range” for the volume and concentration of droplets.
For example, in the preparation of samples of plasmid DNA, the pH of the
solution must be less than 8.0 to avoid a significant reduction in the efficiency
of the lysozyme [
10
]. If an unexpected error occurs during the experiment or
the requirements of the bioassay protocol are violated, the outcome of the entire
experiment are incorrect. When this occurs, all the steps of the experiment must be
repeated to correct the error [
11
,
12
]. Repetition of experiments leads to wastage of
expensive reagents and hard-to-obtain samples.
The repetitive execution of on-chip laboratory experiments results in the follow-
ing problems: (i) an increase in the time-to-result for a bioassay, which is detrimental
to real-time detection and rapid response; (ii) wastage of samples that are difficult
to obtain or prepare, as well as the wastage of expensive reagents.
Therefore, it is necessary to develop techniques for monitoring assay outcomes
at intermediate stages and design an efficient error-recovery mechanism. Error
recovery in digital microfluidics has received relatively little attention in the
literature. The only reported work is [
11
], which proposed intermediate stage
monitoring and rollback error-recovery for a microfluidic biochip. In [
11
], sensing
system is used to verify the correctness of immediate product droplets at various
steps in the on-chip experiment. When an error is detected at a sensor, i.e., the
volume or concentration of the droplet is below or above the acceptable calibrated
range, the corresponding droplet is discarded. If the outputs of an operation fail
to meet the quality requirements that are derived based on sensor calibration, the
operation will be re-executed. New product droplets will be generated to replace the
unqualified droplet.
Figure
2.1
shows an example of rollback error-recovery. The initial sequencing
graph of a bioassay is shown in Fig.
2.1
a. Here we assume that the outputs of each
dispensing, mixing and splitting operation are evaluated by a sensor. When an error
occurs at operation 9, the system will re-execute the corresponding dispensing and
mixing operations. Figure
2.1
b shows the new sequencing graph for error-recovery,
where operations 12, 13, and 14 are added to generate new product droplets. In
the absence of “physical-aware” control software, the error-recovery method in [
11
]
suffers from following drawbacks:
