Combined screening criteria cut estimated errors in rat nerve injury study
Combining two screening criteria reduced the estimated error rate for detecting changes in gene activity after nerve injury in rats from 16% to 5%. The study, published in BMC Neurosci on 25 October 2002, examined the dorsal root ganglion, a cluster of nerve cell bodies, 3 days after the sciatic nerve was cut. The combined screen identified 240 potentially regulated genes, compared with 178 found using a 2-fold change threshold alone. Their functions suggested changes in cell structure, activation of genes expressed by immune and inflammatory cells, and reduced activity of genes involved in nerve signalling.
The researchers used oligonucleotide microarrays, tools that measure the activity of many genes at once. Each of two comparisons used a set of three replicate arrays. One compared uninjured rats with other uninjured rats to assess measurement variability. The other compared uninjured rats with rats whose sciatic nerve had been cut.
The variability results supported using P < 0.05, a statistical threshold for judging whether a measured difference reflects more than chance, despite the large number of tests. Combining that threshold with a >1.5-fold change in gene activity produced the lower estimated error rate. It also flagged more candidate genes than the 2-fold threshold alone. Many had not previously been described as regulated in the dorsal root ganglion after nerve cutting.
The team checked 24 transcripts, the RNA messages produced when genes are active, using Northern blot, quantitative slot blots and in situ hybridization. These laboratory checks agreed with the arrays 83% of the time. Most mismatches involved genes with low activity, reflecting the arrays' limited sensitivity at those levels.
Measured differences in messenger RNA also correlated with the relative changes recorded by the arrays, with r2 = 0.8567. Individual genes showed different patterns of regulation over time.