SUMMARY CDK4/6 inhibitors (CDK4/6i) are effective in breast cancer, however drug resistance is frequently encountered and poorly understood. We conducted a genomic analysis of 348 estrogen receptor-positive breast cancers treated with CDK4/6i and identified loss of function mutations affecting FAT1 and RB1 linked to drug resistance. FAT1 loss led to marked elevations in CDK6 whose suppression restored sensitivity to CDK4/6i. The induction of CDK6 was mediated by the Hippo pathway with accumulation of YAP and TAZ transcription factors on the CDK6 promoter. Genomic alterations in other Hippo pathway components were also found to promote CDK4/6i resistance. These findings uncover a tumor suppressor function of Hippo signaling in ER+ breast cancer and establish FAT1 loss as a mechanism of resistance to CDK4/6i.

Bigtable is a distributed storage system for managing structured data that is designed to scale to a very large size: petabytes of data across thousands of commodity servers. Many projects at Google store data in Bigtable, including web indexing, Google Earth, and Google Finance. These applications place very different demands on Bigtable, both in terms of data size (from URLs to web pages to satellite imagery) and latency requirements (from backend bulk processing to real-time data serving). Despite these varied demands, Bigtable has successfully provided a flexible, high-performance solution for all of these Google products. In this article, we describe the simple data model provided by Bigtable, which gives clients dynamic control over data layout and format, and we describe the design and implementation of Bigtable.

No abstract

Single-language runtime systems, in the form of Java virtual machines, are widely deployed platforms for executing untrusted mobile code. These runtimes provide some of the features that operating systems provide: inter-application memory protection and basic system services. They do not, however, provide the ability to isolate applications from each other, or limit their resource consumption. This paper describes KaffeOS, a system that provides these features for a Java runtime. The KaffeOS architecture takes many lessons from operating system design, such as the use of a user/kernel boundary.The KaffeOS architecture supports the OS abstraction of a process in a Java virtual machine. Each process executes as if it were run in its own virtual machine, including separate garbage collection of its own heap. The difficulty in designing KaffeOS lay in balancing the goals of isolation and resource management against the goal of allowing direct sharing. Overall, KaffeOS is up to 11% slower than the freely available JVM on which it is based, which is an acceptable penalty for the safety that it provides. KaffeOS is substantially slower than commercial JVMs, but exhibits much better performance scaling in the presence of uncooperative code.

Dynamic code generation allows programmers t o u s e r un-time information in order to achieve performance and expressiveness superior to those of static code. The`C Tick C language is a superset of ANSI C that supports e cient and high-level use of dynamic code generation.`C provides dynamic code generation at the level of C expressions and statements, and supports the composition of dynamic c o d e a t r un time. These features enable programmers to add dynamic code generation to existing C code incrementally, and to write important applications such a s just-in-time" compilers easily. The paper presents many examples of hoẁ C can be used to solve practical problems.The tcc compiler is an e cient, portable, and freely a vailable implementation of`C. tcc allows programmers to trade dynamic c o m pilation speed for dynamic c o d e q u a l ity: in some applications, it is most important to generate code quickly, while in others code quality m atters m ore than compilation speed. The overhead of dynamic c o m pilation is on the order of 100 to 600 cycles per generated instruction, depending on the level of dynamic optimization. Measurements show that the use of dynamic code generation can improve p e r formance by a l most a n o r d e r o f m agnitude; two-to four-fold speedups are common. In most cases, the overhead of dynamic compilation is recovered in under 100 uses of the dynamic code; sometimes it can be recovered within one use. C is a superset of ANSI C that supports the high-level and e cient use of dynamic code generation. It extends ANSI C with a small number of constructs that allow the programmer to express dynamic code at the level of C e xpressions and statements, a nd to compose arbitrary dynamiccode at r un time. These features enable programmers to write complex imperative c ode manipulation programs in a style similar to Lisp Steele Jr. 1990 , and make it relatively easy to write powerful and portable dynamic code. Furthermore, since`C is a superset of ANSI C, it is not di cult to improve p e r formance of code incrementally by a dding dynamic code generation t o e xisting C programs.C's extension s t o C | t wo t ype constructors, three unary o perators, a nd a few special forms | allow dynamic code to betype-checked statically. Much of the overhead of dynamic compilation can therefore be incurred statically, which improves the e ciency of d y namic compilation. While these constructs were designed for ANSI C, it should be straightforward to add analogous constructs to other statically typed languages.tcc is an e cient a nd freely available implementation of`C, consisting of a f r ont end, back ends that compile to C and to MIPS and SPARC assembly, and two runtime systems. tcc allows the user to trade dynamic code quality for dynamic code generation speed. If compilation speed must bemaximized, dynamic code generation and register allocation can be performed in one pass; if code quality is most important, the system can construct and optimize an intermediate representation prior to code genera...