A study of over 100 ancient Roman genomes reveals the empire's role as a genetic crossroads for people throughout Europe and the Mediterranean. An international team of scientists analyzed 127 genomes from 29 sites in and around Rome spanning 12,000 years. As reported in Science this week, they find two major prehistoric ancestry transitions: one occurring with the introduction of farming around 7,000 years ago and another prior to the Iron Age. By the time Rome was founded in 753 BC, the genetic composition of the region was close to that of the modern Mediterranean. During the Imperial through Medieval periods, however, Rome experienced growth in genetic diversity, reflecting contributions from individuals coming from the Near East, Europe, and North Africa, and coinciding with major historical events. Our sister publication GenomeWeb has more on the study here.
Using a CRISPR-based screening approach, a Massachusetts Institute of Technology-led team has identified genetic vulnerabilities in small cell lung cancer (SCLC), an aggressive lung cancer subtype lacking targetable genetic drivers. In their study, which appears in Science Translational Medicine, the investigators used a single-guide RNA library targeting roughly 5,000 genes encoding druggable proteins to perform loss-of-function genetic screens in a panel of cell lines derived from mouse models of SCLC, lung adenocarcinoma, and pancreatic ductal adenocarcinoma. Cross-cancer analyses, they write, enabled the identification of SCLC-specific targets, most notably the pyrimidine biosynthesis pathway. When a key enzyme in this pathway was inhibited with an agent called brequinar, SCLC tumor growth was suppressed and survival was increased in disease models.
A new machine-learning method has been shown to accurately predict genome modification activity of the CRISPR enzyme SpCas9. In a report appearing in Science Advances, a team of Korean scientists describe training the computational model — dubbed DeepSpCas9 — on a large dataset of SpCas9-induced indel frequencies, and then testing it against a variety of independent datasets in the literature. They find it had high generalization performance, and fared better than nine other previously reported models.