6. Interpreting the Scanned Image

The end product of a comparative hybridization experiment is a scanned array image. A small piece of such an image is shown above. The measured intensities from the two fluorescent reporters have been false-colored red and green and overlaid. Yellow spots have roughly equal amounts of bound cDNA from each cell population and so have equal intensity in the red and green channels (red green = yellow). Spots whose mRNA's are present at a higher level in one or the other cell population show up as predominantly red or green.

The intensities provided by the array image can be quantified by measuring the average or integrated intensities of the spots. The ratio of fluorescent intensities for a spot is interpreted as the ratio of concentrations for its corresponding mRNA in the two cell populations. Schena et al (1996) have demonstrated the ability to detect quantitative changes of as little as a factor of two, with reasonable agreement between expression ratios measured on the array and ratios measured by an alternate form of RNA blotting.

Interpreting the data from a microarray experiment can be challenging. Quantitation of the intensities on each spot is subject to noise from irregular spots, dust on the slide, and nonspecific hybridization. Deciding the intensity threshold between spots and background can be difficult, especially when the spots fade gradually around their edges. Detection efficiency might not be uniform across the slide, leading to excessive red intensity on one side of the array and excessive green on the other side. Even after overcoming detection and calibration problems, the measured intensities for each spot only represent the ratio of cDNA's in each cell population. Low levels of cDNA due to reverse transcription bias, sample loss, or an inherently rare mRNA can cause large uncertainties in these ratios.

Numerous software packages, both free and commercial, exist for quantitating microarray data. I have developed one such program, Dapple, to address some of the abovementioned image quality issues, in the hope that its methods might be integrated into other array quantitation systems.

Typically, the interpreted array data will highlight a relatively small number of spots representing very differentially-expressed mRNA's whose genes deserve further investigation. Alternatively, the overall pattern of expression can be used as a "fingerprint" to characterize specific cell types (e.g. different classes of tumors), even if not all the differentially-expressed genes on an array have been identified.

Notes

[1] The rest is ribosomal RNA (rRNA) and transfer RNA (tRNA).

[2] Predecessors to current microarray technology added radioactive phosphorus to the cDNA molecules, so that hybridized cDNA's would form spots on X-ray film (or more sensitive phosphorimaging devices). This labeling technology is inadequate for comparative hybridization to the same microarray, since we have to distinguish the two different samples from each other.