Twenty years of technology.

In looking back at previous issues of RNA and my own eight RNA papers, what strikes me, in addition to all the great science that’s been published in our society journal over the past 20 years, is the evolution (or more properly, revolution!) in the methods and technologies we use. Equally striking has been the exponential increase in the amount of data in each published paper. In two decades, post-transcriptional gene regulation has evolved from a mostly qualitative science into a highly quantitative one. We’ve gone from spending most of our time just trying to get one experiment at the bench to work to now spending months puzzling overmassive datasets generated by high throughput instruments in research cores. As I look to the future, I wonder about the technologies yet to come, howwe can possibly deal with the ever increasing size of our datasets, and what will constitute a story “complete enough” to warrant publication 20 years from now. When I entered graduate school in 1984, Sanger sequencing on large format gels and site-directed mutagenesis using the single-stranded M13 bacteriophage were state-of-the-art. Oligonucleotide synthesis was so time-consuming and expensive that ordering the wrong 17 nucleotide DNA oligo was a major catastrophe. Whole PhD dissertations were devoted to the sequencing of one gene, or the generation and characterization of just a few site-specific point mutations. By the time I arrived in 1989 as a postdoc in Phil Sharp’s lab, PCR was the next new thing. I remember how our first PCR machine, shared with the adjoining Houseman lab (∼35 researchers in total), was in such high demand that major yelling matches ensued when samples were left in a minute too long. By that time runoff transcription and in vitro splicing assays were well established, and native gels and density gradients had revealed that the pre-mRNA splicingmachinery was exceptionally large and complex.Mymain contribution in that period was to show that T4 DNA ligase could join together long RNA molecules, enabling the synthesis of long, site-specifically modified RNAs. Eric Sontheimer in Joan Steitz’s group immediately started using this method to incorporate 4-thio-U at key sites in premRNAs, ushering in a decade of site-specific photocrosslinking experiments. By the time I started my own lab at Brandeis in 1994, the age of genomics was already in full swing. With Craig Venter’s 1991 publication of >170,000 expressed sequence tags (ESTs), sequenced on first generation of automated machines, came the realization that alternative splicing was much more prevalent than had previously been thought. Soon serial analysis of gene expression (SAGE; Kinzler 1995) and high-density microarrays (Brown 1995) provided tools to quantitatively measure how the abundance of thousands of mRNAs at once responded to changing cell conditions. Of course substantial completion of the human genome at the turn of the newmillenniumwas another major milestone, with Haussler’s UCSC genome browser enabling researchers to easily visualize the gDNA organization of favorite genes. The emergence of RNA-Seq (Wold 2008) coincided almost exactly with my move to UMass Medical School in late 2007. Having theretofore sworn off high-throughput methods in my own lab after an incredibly frustrating attempt to build a microarray facility at Brandeis a decade prior, my lab began making its first deep sequencing libraries within a year of my arrival at UMMS. In our case, the focus has been on mapping transcriptome-wide sites of RNP complex occupancy. Akin to ribosome profiling (Ingolia/ Weissman, 2011), RNP footprinting reveals the RNA binding sites of multicomponent complexes of defined protein composition (e.g., the exon junction complex, EJC, or lariat intron-containing spliceosomes). Complementary methods developed a few years before map transcriptome-wide sites of direct RNA–protein contact (e.g., HITS-CLIP, Darnell 2008; PAR-CLIP, Tuschl 2010). So in just 20 years, we’ve gone from mapping one RNA–protein crosslink at a time to mapping literally millions in one shot. Major challenges for the future are how to determine which of these millions are of functional relevance and how best to integrate the hundreds of currentlyand soon-to-be-available datasets to extract even more meaning. Another area of great technological innovation over the last 20 years has been quantitative proteomics. Major innovations in the 1990’s made it possible to deconvolve the spectral mess produced by trypsin digestion of complex mixtures. In collaboration with Steve Gigi, we published in RNA in 2002