Along those same lines, researchers have also run into challenges translating standard data normalization methods for bulk nucleic acid preparations to single-cell data.
For instance, using “housekeeping” genes to normalize qPCR expression data does not make sense at the single-cell level.
In addition to expression within single cells, Dr. Flynn is interested in studying variations in expression among cells. Being able to discern technical variance from biological variance is a major concern for many of the amplification strategies prior to assay measurement, Dr. Flynn says.
Dr. Flynn and his colleagues are using large expression datasets from nanofluidic qPCR arrays and homogenous reference samples from entire cell populations to normalize data. This approach is somewhat analogous to microarray formats that normalize the signal from each probe spot to the array itself combined with spiked in controls. Overall, this strategy has been fairly successful in validating the single-cell measurements.
Another concern when analyzing single-cell data is that many standard statistical approaches are unusable because much of the single-cell data violates the basic assumptions within these tests.
“The all-too popular Student’s t-test is not appropriate for gene expression comparisons at the single-cell level. We are taking the stance that data analysis must be non-parametric and should be the new standard in single-cell analysis,” Dr. Flynn says.
Data interpretation can indeed be difficult. According to Cincinnati Children’s Hospital’s Dr. Potter, the limitations relate to existence of fewer than ten transcripts per expressed gene, and gene expression occurring in a burst mode, not as a steady state process. “We have made much progress, but there is still considerable room for improvement,” he says.
As Dr. Flynn at the Buck Institute puts it: “What I think will be of broad interest to the biotech community is how single-cell biology will change our approach to the development of disease treatments,” he says. “What is really an unknown at the moment is how diseases progress on a cell-by-cell basis.”
While it is established that cancers can begin with a single cell escaping cell cycle control, it is not so clear if the progression of other maladies come from similar stochastic changes. Single-cell analysis is a potent tool in identifying not only the disease pathogenesis, but also in the development of targeted cell therapies. Current approaches are based on highly sensitive single-cell tests to diagnose diseases prior to the development of clinical symptoms.
Further, single-cell analysis allows for sample sizes of hundreds if not thousands of individual cells in order to detect disease in a normal cell haystack.
Overall, single-cell technology will drive both the sensitivity and specificity of future clinical assays up, Dr. Flynn says.