Let's consider the scenario where we would like to select genes from a differential expression analysis list that show at least 30% increase or 30% decrease in expression. This can be confusing at first if you attempt it based on the log fold change values. So let's look into this further -
Case 1 - 30% Increase:
Let's assume the expression of gene X is 10 in the control condition and 13 in the treatment condition. In this case, gene X is up-regulated and there is a 30% increase in expression (13-10)/10. In this case, the foldChange would be 13/10 = 1.3. And the log2FoldChange would be log2(13/10) = 0.378. Therefore, one may use a log2FoldChange threshold of 0.378 and above to obtain genes that are up-regulated with at least 30% increase.
Case 2 - 30% Decrease:
Assuming the expression of gene X is 10 in the control condition and 7 in the treatment condition, gene X is down-regulated and there is a 30% decrease in expression (7-10)/10. In this case, foldChange = 7/10 = 0.7 and the log2FoldChange = log2(7/10) = -0.514. Therefore, one may use a log2FoldChange threshold of -0.514 (and below) to obtain genes that are down-regulated with at least 30% decrease.
Convert log2FoldChange to foldChange:
2^log2FoldChange = foldChange. For example: 2 ^ -0.514 = 0.7