RT Journal Article SR Electronic T1 A systematic review and meta-analysis of Drosophila short-term-memory genetics: robust reproducibility, but little independent replication JF bioRxiv FD Cold Spring Harbor Laboratory SP 247650 DO 10.1101/247650 A1 Tayfun Tumkaya A1 Stanislav Ott A1 Adam Claridge-Chang YR 2018 UL http://biorxiv.org/content/early/2018/01/13/247650.abstract AB Preface We invite memory researchers to submit comments, anonymous or signed, by email, social media, or directly to the manuscript. To read text and figures side-by-side on a desktop, please make a duplicate of this file.Abstract Geneticists have long used olfactory conditioning techniques in Drosophila to identify the neurons and genes that mediate learning. While this method has characterized an abundance of memory-related genes, little is known about how these genes induce short-term memory (STM) via signaling pathways; characterizing these networks will be essential to developing mechanistic models of memory formation. Here, we investigated why elucidating the STM pathways has been relatively slow. One possibility is that the STM evidence base is weak due to publication of poorly reproducible results, as has been observed in other fields. We examined this hypothesis by performing a systematic review and subsequent meta-analysis of the STM genetics field. Using several metrics to quantify the variation between discovery articles and follow-up studies, we found that seven genes were highly replicated, showed no publication bias, and had generally high reproducibility. However, the remaining ~80% memory genes have not been replicated since their initial discovery. Although we observed only a few studies that investigated gene interactions, the reviewed genes could together account for >1000% memory. This large summed effect size indicates either that some of the gene findings are not reproducible, that many memory genes participate in shared pathways, or that current protocols lack the specificity needed to identify core plasticity memory genes. Mechanistic theories of memory and cognition will require the convergence of evidence from system, circuit, cellular, molecular, and genetic experiments. As this study demonstrates, systematic data synthesis is an essential tool for this integrated brain science.