µParaflo® microfluidic array platform of LC Sciences has an excellent assay reproducibility. Assay reproducibility is measured by how small chip-to-chip variation is for the same sample assayed over multiple chips. For each miRNA the assay variation across multiple chips is calculated using variation coefficient (CV):
CVmiR-i = Standard Deviation (ImiR-i) / Average (ImiR-i)
where ImiR-i is the signal intensity of miRNA-i.
Figure 1 shows a histogram of CV from six individual assays using a total RNA sample of human kidney tissue. A CV peak value of 8% and a median value of 11% were measured. These numbers indicate a low system noise that is more than sufficient to detect a 1.5-fold change that is commonly used as a threshold number for indicating significant biological differentiation.
Figure 1. A CV histogram based on six individual assays using a total RNA sample of human kidney tissue. Data are collected from average signal intensities above 32.
µParaflo® microfluidic array platform of LC Sciences produces highly consistent dual-sample assay data and is capable of revealing even very small differences between two samples. The dual-sample consistency is measured by how small signal difference is between Cy3 and Cy5 channels of the same sample source. The signal difference is measured by variation coefficients between Cy3 and Cy5 signals.
CVi = Standard Deviation (ICy3i, ICy5i) / Average (ICy3i, ICy5i),
where ICy3i and ICy5i are signal intensities of Cy3 and Cy5 channels at spot i.
Figure 2a plots CV histogram of a dual sample assay data using a total RNA sample of human muscle tissue. A low CV peak value of 4% and a median value of 6% were measured. Figure 2b is a scatter plot of the Cy3 and Cy5 signals used to calculate the CV values of 2a. A high Pearson coefficient of 0.992 indicates a high consistency between the Cy3 and Cy5 signals.
Figure 2. CV histogram of a dual sample assay data using a total RNA sample of human muscle tissue. Data are collected from average signal intensities above 16.
Figure 3. A scatter plot of the Cy3 and Cy5 signals used to calculate the CV values of 2.
Our microRNA detection dynamic range is no less than 3.5 logs and the lower detection limit is less than 10 attomole. These numbers are derived by using an experimental design method called the Latin Square Test.
In our experiments, we selected and synthesized 10 microRNA transcript sequences that have low expression levels in a cell-line total RNA sample which is used to generate a real background. We prepared 10 samples with each one containing a fixed amount of according to a predetermined spiking concentration matrix. The concentrations of the 10 different spiking microRNAs in each sample cover 3.5 logs. The 10 samples are hybridized to 10 individual chips. The resulting signal intensities of the Latin Square test are plotted in the figure to the left. The data illustrates a dynamic range of no less than 3.5 logs. Signal intensities of spiked-in RNA samples are plotted against spiking concentrations. The signal intensities of each chip were normalized against the signal intensity of a constant spiking control.
The lower detection limit is determined based on a p-value calculation using the signals of Cy3 and Cy5 channels. A p-value cut-off of 0.04 is used.
Even with a single-base mismatch (1MM) in a detection probe most signal drops by more than 30 fold and a perfect match (PM) to 1MM ratio of more than 100 fold is achieved.
The high assay specificity is illustrated in an extensive test for assessing the specificities of all mouse microRNA probes. On a microRNA array chip, for each microRNA transcript, in addition to a perfect match probe, three mismatch probes containing one, two and three substitution mismatches, respectively, were added.
The figure below is a section of the array image showing binding intensities of all four types of probes. The results show the majority of the 1MM binding signals drop by more than 30 fold; all except two of the 2MM and all 3MM signals fall into the background level.
Through the use of the µParaflo® microfluidic chip technology and design of probes containing proprietary chemical modifications, we can achieve very uniform hybridization that is extremely reproducible. Much more reproducible than is possible with a spotted array.
|Spotted Array from a reputable facility||µParaflo® microfluidic biochip|
- These are raw data. Click the images to take a closer look.
- Inset shows enlarged spots and intensity profiles through spots.
- Color scheme: blue (0) to red (65000).
- Intensity within each feature varies 3X (from 5000 to 15000) on the spotted array.
- Intensity variation within each feature is negligible on the µParaflo® array.
The figure below illustrates the intra- and inter-chip reliability of our µParaflo® microarrays.
We performed a “color reversal” experiment using mouse brain and mouse thymus RNA samples purchased from Ambion. The results are compared with QPCR data for the same two RNA samples, also purchased from Ambion, published by Applied Biosystems (ABI)*. The comparison data includes all 12 microRNA transcripts published by ABI without any selective picking of the transcripts. As shown in the bar graphs below, the relative intensities of different microRNA transcripts of our array data are in excellent agreement with ABI’s QPCR results.
What is Color Reversal?
“Color reversal” involves two chips. On the first chip sample “A” is labeled with Cy5 and sample “B” with Cy3. On the second chip sample “A” is labeled with Cy3 and “B” with Cy5. The first chip gives a signal ratio of A/B as Cy5/Cy3; the second chip gives A/B as Cy3/Cy5. By correlating the results from the two chips most of the labeling, handling, and system related biases can be eliminated and therefore calls can be narrowed down to true biological differences.