Expression values were normalized within study (see Normalization for more details). For this plot, expression values were further scaled to a median value of 100 within each individual study to facilitate cross study comparisons. The boxplots display the minimum, 25% quantile, median, 75% quantile, and maximum value for each probe that maps to the given gene. Probes are sorted by median expression values. The shaded regions represent low, medium, high, and very high expression ranges (ranges determined by including 40% of genes in the low range, 30% in the middle range, 20% in the high range, and 10% in the top range). Blue boxplots are probes from Array collection studies and gray boxplots are probes from Consortium collection studies.

The plot displays the fraction of samples that were determined to be present for each study/probe (see Normalization for more details). All probes that map to the given gene are displayed, sorted by the probe’s median expression value.

The plots display the fold changes relative to control for each of three diseases (bipolar, schizophrenia, and depression) for each study/probe. 95% confidence intervals are also displayed. Consensus fold change values are also shown, using either all probes, just probes from Array collection studies, or just probes from Consortium collection studies. The consensus fold changes and associated 95% confidence intervals are based on a weighted combination of the individual fold changes and standard errors. The weights are equal to 1/SE_i, where SE_i is the standard error for the i’th probe for the gene. P-values are also shown for the consensus fold changes, and are calculated based on the same weighted fold changes and standard errors used to construct the confidence intervals.

The plots show consensus fold change values for each demographic variable. The consensus fold changes and associated 99% confidence intervals are based on a weighted combination of the individual fold changes and standard errors. The weights are equal to 1/SE_i, where SE_i is the standard error for the i’th probe for the gene. P-values are also shown for the consensus fold changes, and are calculated based on the same weighted fold changes and standard errors used to construct the confidence intervals.