RNA-seq Showdown

by Kelly Rae Chi – Biotechniques

Five RNA-seq library preparation methods go head-to-head in terms of performance in low-quality and low quantity RNA samples. Which method came out on top? 

In a side-by-side comparison of five different transciptome sequencing (RNA-seq) library preparation methods, the RNase H technique outperformed others for analysis of low-quality RNA samples and was among the least expensive; SMART and NuGEN methods worked well for small amounts of RNA. The findings are published online May 19, 2013 in Nature Methods (1).

“We’d been looking for a way to deal with the issue of degraded RNA for a while. We tried different methods, and finally we found this RNase H method that works well and gives consistent results,” said co-author Xian Adiconis, senior research associate at the Broad Institute of Harvard and the Massachusetts Institute of Technology (MIT) in Cambridge, MA.

A strategy for analyzing the transcriptome, RNA-seq uses high-throughput sequencing to sequence and quantify RNA. But results suffer when researchers use samples that are degraded or present in small amounts. Numerous methods and commercial kits are available for the analysis of low quality and low quantity samples, but until now, the choice relied somewhat on guesswork, said Adiconis.

Testing the ability of the techniques to deplete ribosomal RNA (rRNA)—which many scientists hope to eliminate in their RNA-seq experiments—in a low-quality sample, the team found that RNase H eliminated almost all rRNA from a low-quality sample, leaving only .1% of the original amount behind. Meanwhile, the Ribo-Zero method left behind 11.3%, and the NuGEN method left behind 23.2% — though these values vary depending on the extent of RNA degradation. In addition, for low-quality RNA, RNase H and Ribo-Zero libraries more closely correlated with total RNA control libraries and provided more even coverage compared to libraries created with the other methods.

The team then compared RNase H and Ribo-Zero using real samples: a kidney that was formalin fixed and paraffin embedded, and a pancreas sample that was degraded during isolation. In the end, the scientists found that RNaseH outperformed Ribo-Zero and provided a slightly more even coverage.

Overall, RNaseH and Ribo-Zero outperformed the other methods because of a fundamental difference in the way that they specifically target rRNA sequences for depletion. The RNaseH method, for example, uses pre-designed DNA oligonucleotides that bind to ribosomal RNA sequence, and then the RNase H enzyme digests the RNA that is bound to DNA. “This targeted approach leaves the rest of the transcriptome relatively unaffected,” said co-author Rahul Satija, a postdoctoral associate at Broad.

In contrast, other methods, such as polyA selection, DSN, and SMART, indirectly deplete ribosomal RNA by selecting for poly-adenylated mRNA or depleting highly abundant species. The techniques deplete ribosomal RNA but can also introduce bias in the coverage of degraded samples.

In addition, the per-sample costs of the commercial kits for Ribo-Zero, NuGEN, and SMART were found to be substantially higher than those of the other methods such as DSN-lite and RNase H. The amount of labor required to assemble these libraries was about the same, with the exception of DSN-lite which requires an additional day.

Although RNase H seemed to be a clear winner, the choice of a method depends on the research goal. For example, NuGEN works exceptionally well on very small samples. “What we tested was 1 ng and 100 ng input, but actually [NuGEN methods] can go much lower than that. They can work on already degraded RNA as well,” Adiconis said.

SMART methods also work well on tiny samples—though the RNA must be intact. For another new study published in Nature, Adiconis and others from the same team used SMART methods to sequence the transcriptome of individual immune cells, showing the heterogeneity of these single cells in their gene expression (2).

NuGEN has worked for the group in the past, to measure the transcriptomes in ultra-low input viral samples–RNA in less than 1 picogram amounts. “[But] we’ve never tried the NuGEN method for single cells, so we can’t say that it’s better,” said Joshua Levin, of the Genome Sequencing and Analysis Program (GSAP) at Broad, who co-led the Nature Methods study and co-authored the Nature study.

Two additional RNA-seq methods were recently published but not in time to make it into the group’s comparison (3).

1. Adiconis, X., D. Borges-Rivera, R. Satija, D. S. DeLuca, M. A. Busby, A. M. Berlin, A. Sivachenko, D. A. Thompson, A. Wysoker, T. Fennell, A. Gnirke, N. Pochet, A. Regev, and J.Z. Levin. 2013. Comparative analysis of RNA sequencing methods for degraded or low-input samples. Nature Methods (May). [abstract]

2. Shalek, A. K., R. Satija, X. Adiconis, R. S. Gertner, J. T. Gaublomme, R. Raychowdhury, S. Schwartz, N. Yosef, C. Malboeuf, D. Lu, J. T. Trombetta, D. Gennert, A. Gnirke, A. Goren, N. Hacohen, J. Z. Levin, H. Park, and A. Regev. 2013. Single-cell transcriptomics reveals bimodality in expression and splicing in immune cells. Nature (May). [abstract]

3. Pan, X. et al. 2013. Two methods for full-length RNA sequencing for low quantities of cells and single cells. Proc. Natl. Acad. Sci. USA 11110:594–599. [abstract]