The P-adjusted value is preferred in most cases because it accounts for the False Discovery Rate, or false positives. In some cases, the P-value can be used: for example, if there are few samples in the groups being compared, or if not enough genes are shown to be differentially expressed when using the P-adjusted value.