RNA-seq experiments should best be carried out with samples of consistent RNA integrity and input amounts.  However, some RNA-seq samples can be so limited and irreplaceable that experiments have to be carried out with less than the recommended input amounts. Similar complications can occur if some of the samples are significantly more degraded than others. Such situations require weighing the pros and cons when choosing the input amounts from the more abundant samples. Points to consider are:

• The ideal approach for an RNA-seq project would be to treat each sample exactly the same, to minimize technically induced variation in the resulting data.  This would include starting each library prep with the same amount of total RNA input and applying the same number of PCR cycles to each of the libraries. However, more degraded samples usually require increased input amounts.
• In general, sequencing library preparations do not fail at a specific input amount threshold. Lower amounts can usually be compensated for by increasing the number of PCR cycles during the preparation. Thus, inputs lower than the kit manufacturers' recommendations can be used in some cases. Any reduced input amounts (and/or higher sample degradation) will, however, lead to reduced library complexities, and thus noisier gene expression data. The best data are usually generated when working with input amounts in the upper half of the manufacturers' input recommendations.

For sample sets with varying RNA sample amounts and qualities, we suggest verifying first if outlier samples with significantly lower sample amounts or lower quality can be dropped from the experiment. If this is not the case we suggest two options. We will ask you to pick one of these or to provide detailed instructions for another approach: Strategy #1:  Normalize all RNA input amounts to the lowest mass sample that has to be included in the study.  Please note that this will more severely impact the quality for the originally high RNA quality & high RNA amount samples. Strategy #2: Normalize the RNA input amounts to a range from the lowest input sample to three times that of the lowest input sample. With this approach, all libraries will still undergo the same number of PCR cycles, which preserves more of the sample quality of the more abundant and higher quality samples. (An example case would be that the lowest available amount for one of the samples is 10 ng. We would then dilute only high-amount samples to an input of at most 30 ng.) For most projects, we tend to recommend Strategy #2, especially if the ratio of the low-input outlier samples is low.