Search

We hypothesized that small molecule transcriptional perturbation could be harnessed to target a cellular dependency involving protein arginine methyltransferase 5 (PRMT5) in the context of methylthioadenosine phosphorylase (MTAP) deletion, seen frequently in malignant pleural mesothelioma (MPM). Here we show, that MTAP deletion is negatively prognostic in MPM. In vitro, the off-patent antibiotic Quinacrine efficiently suppressed PRMT5 transcription, causing chromatin remodelling with reduced global histone H4 symmetrical demethylation. Quinacrine phenocopied PRMT5 RNA interference and small molecule PRMT5 inhibition, reducing clonogenicity in an MTAP-dependent manner. This activity required a functional PRMT5 methyltransferase as MTAP negative cells were rescued by exogenous wild type PRMT5, but not a PRMT5E444Q methyltransferase-dead mutant. We identified c-jun as an essential PRMT5 transcription factor and a probable target for Quinacrine. Our results therefore suggest that small molecule-based transcriptional perturbation of PRMT5 can leverage a mutation-selective vulnerability, that is therapeutically tractable, and has relevance to 9p21 deleted cancers including MPM.

We hypothesized that small molecule transcriptional perturbation could be harnessed to target a cellular dependency involving protein arginine methyltransferase 5 (PRMT5) in the context of methylthioadenosine phosphorylase (MTAP) deletion, seen frequently in malignant pleural mesothelioma (MPM). Here we show, that MTAP deletion is negatively prognostic in MPM. In vitro, the off-patent antibiotic Quinacrine efficiently suppressed PRMT5 transcription, causing chromatin remodelling with reduced global histone H4 symmetrical demethylation. Quinacrine phenocopied PRMT5 RNA interference and small molecule PRMT5 inhibition, reducing clonogenicity in an MTAP-dependent manner. This activity required a functional PRMT5 methyltransferase as MTAP negative cells were rescued by exogenous wild type PRMT5, but not a PRMT5E444Q methyltransferase-dead mutant. We identified c-jun as an essential PRMT5 transcription factor and a probable target for Quinacrine. Our results therefore suggest that small molecule-based transcriptional perturbation of PRMT5 can leverage a mutation-selective vulnerability, that is therapeutically tractable, and has relevance to 9p21 deleted cancers including MPM.

Scientific communities are increasingly adopting machine learning and deep learning models in their applications to accelerate scientific insights. High performance computing systems are pushing the frontiers of performance with a rich diversity of hardware resources and massive scale-out capabilities. There is a critical need to understand fair and effective benchmarking of machine learning applications that are representative of real-world scientific use cases. MLPerf is a community-driven standard to benchmark machine learning workloads, focusing on end-to-end performance metrics. In this paper, we introduce MLPerf HPC, a benchmark suite of large-scale scientific machine learning training applications driven by the MLCommons Association. We present the results from the first submission round, including a diverse set of some of the world's largest HPC systems. We develop a systematic framework for their joint analysis and compare them in terms of data staging, algorithmic convergence, and compute performance. As a result, we gain a quantitative understanding of optimizations on different subsystems such as staging and on-node loading of data, compute-unit utilization, and communication scheduling, enabling overall $>10 \times$ (end-to-end) performance improvements through system scaling. Notably, our analysis shows a scale-dependent interplay between the dataset size, a system's memory hierarchy, and training convergence that underlines the importance of near-compute storage. To overcome the data-parallel scalability challenge at large batch sizes, we discuss specific learning techniques and hybrid data-and-model parallelism that are effective on large systems. We conclude by characterizing each benchmark with respect to low-level memory, I/O, and network behavior to parameterize extended roofline performance models in future rounds.

This paper presents the first industry-standard open-source machine learning (ML) benchmark to allow perfor mance and accuracy evaluation of mobile devices with different AI chips and software stacks. The benchmark draws from the expertise of leading mobile-SoC vendors, ML-framework providers, and model producers. It comprises a suite of models that operate with standard data sets, quality metrics and run rules. We describe the design and implementation of this domain-specific ML benchmark. The current benchmark version comes as a mobile app for different computer vision and natural language processing tasks. The benchmark also supports non-smartphone devices, such as laptops and mobile PCs. Benchmark results from the first two rounds reveal the overwhelming complexity of the underlying mobile ML system stack, emphasizing the need for transparency in mobile ML performance analysis. The results also show that the strides being made all through the ML stack improve performance. Within six months, offline throughput improved by 3x, while latency reduced by as much as 12x. ML is an evolving field with changing use cases, models, data sets and quality targets. MLPerf Mobile will evolve and serve as an open-source community framework to guide research and innovation for mobile AI.

Machine-learning (ML) hardware and software system demand is burgeoning. Driven by ML applications, the number of different ML inference systems has exploded. Over 100 organizations are building ML inference chips, and the systems that incorporate existing models span at least three orders of magnitude in power consumption and five orders of magnitude in performance; they range from embedded devices to data-center solutions. Fueling the hardware are a dozen or more software frameworks and libraries. The myriad combinations of ML hardware and ML software make assessing ML-system performance in an architecture-neutral, representative, and reproducible manner challenging. There is a clear need for industry-wide standard ML benchmarking and evaluation criteria. MLPerf Inference answers that call. In this paper, we present our benchmarking method for evaluating ML inference systems. Driven by more than 30 organizations as well as more than 200 ML engineers and practitioners, MLPerf prescribes a set of rules and best practices to ensure comparability across systems with wildly differing architectures. The first call for submissions garnered more than 600 reproducible inference-performance measurements from 14 organizations, representing over 30 systems that showcase a wide range of capabilities. The submissions attest to the benchmark's flexibility and adaptability., Comment: ISCA 2020

Machine learning (ML) needs industry-standard performance benchmarks to support design and competitive evaluation of the many emerging software and hardware solutions for ML. But ML training presents three unique benchmarking challenges absent from other domains: optimizations that improve training throughput can increase the time to solution, training is stochastic and time to solution exhibits high variance, and software and hardware systems are so diverse that fair benchmarking with the same binary, code, and even hyperparameters is difficult. We therefore present MLPerf, an ML benchmark that overcomes these challenges. Our analysis quantitatively evaluates MLPerf's efficacy at driving performance and scalability improvements across two rounds of results from multiple vendors., Comment: MLSys 2020

Background: DREAM is an open-label, single arm, multi-centre, phase II trial with a safety run in designed to determine the activity, safety and tolerability of durvalumab combined with cisplatin and pemetrexed as first line therapy in malignant pleural mesothelioma (MPM). ANZ Clinical trial registry number: ACTRN12616001170415. Methods: Chemotherapy-naïve patients with MPM of all histological subtypes, planned for first-line cisplatin and pemetrexed who had not received prior radiotherapy and had an Eastern Cooperative Oncology Group (ECOG) performance status of 0-1 were eligible. Patients received durvalumab (1125mg), cisplatin (75mg/m2) and pemetrexed (500mg/m2) 3-weekly for a maximum of 6 cycles, followed by durvalumab alone (1125 mg 3-weekly) until progression or to 12 months total therapy. Primary endpoint was progression-free survival at 6 months (PFS6). The design provides >90% power if the true PFS6 rate is 65%, with 5% type I error using an expected PFS6 from standard treatment of 45%. Matching tissue and bloods were collected for translational research. Results: Between Dec 2016 to Sep 2017 54 participants were recruited. The median age was 68 (42-82) with 82% male and 60% ECOG 0. They received a median of 6 cycles of chemotherapy (range 1-6) and 7 of durvalumab (given with and without chemotherapy) (range 1-18). The 6-patient safety run in confirmed the tolerability of chemoimmunotherapy. In the first 31 evaluable patients (pre-specified primary interim analysis), PFS6 based on modified RECIST (mRECIST) for MPM was 71%. The objective response rate (ORR) was 61% using mRECIST and 53% using iRECIST. Thirty-one patients (57%) experienced adverse events (AEs) that were grade ≥3. Nineteen patients (35%) had immune-related (ir) AEs including 2 grade 3 irAEs. Conclusions: In MPM, adding durvalumab to first line treatment with cisplatin and pemetrexed had higher rates of PFS6 and ORR than expected for chemotherapy alone, with acceptable tolerability. The final r

Complex relationships exist between the physical world and the perceptual world. In the auditory domain, the impact of stimulus characteristics on the perception of sound has received a wealth of attention in the literature over the last 60 years. This thesis examined differences in stimulus familiarity, the discrimination of concurrent pitch information and the experience of sensory dissonance. There is still no consensus among researchers regarding the mechanisms responsible for accurate pitch perception. The principle aim of this thesis was to explore the role of training in pitch processing for a range of musical sounds varying in familiarity, as outlined in a new model of sound perception (McLachlan & Wilson, 2010). An additional aim of this thesis was to investigate the perception of dissonance in musicians with varying types of musical training and pitch identification ability. This thesis was divided into two main parts as a means of exploring these aims. Four general hypotheses and nine specific hypotheses were postulated in Part I. The first general hypothesis predicted that pitch matching performance would be better for musicians trained on inharmonic Gamelan instruments compared to musicians with similar levels of training who have not been exposed to these instrumental timbres. It was predicted that Gamelan musicians would rate Gamelan instruments as more familiar than Western musicians in the absence of musical context, and that Gamelan musicians would be better at matching the pitch of inharmonic Gamelan instruments than musicians with similar levels of musical training on Western instruments. The second general hypothesis proposed that trained musicians would exhibit higher levels of pitch matching accuracy for intervals and chords that are more familiar. It was specifically hypothesised that highly trained musicians would rate intervals and chords as higher in familiarity than musicians with less training. It was also predicted that pitch matching p

The synthesis and biological activity of two types of enkephalin analogs are reported: those in which the C-terminal carboxylic acid group was replaced by a chloromethyl ketone, and those in which the C-terminus was replaced by lysine, or a derivative of lysine. The substitution of the terminal carboxyl group of leucine enkephalin and D-Ala('2)-leucine enkephalin with a chloromethyl ketone uniformly increased the potency in the guinea pig ileum assay by a factor of six. The chloromethyl ketones were also more potent in inhibiting the stereospecific binding of {('3)H}-naloxone to opiate receptors in the rat brain. While the chloromethyl ketone analogs had the potential to act as irreversible agonists it was not clear from the present studies that such binding did indeed occur. The increased activity of these derivatives was related to the increased lipophilic character of the peptide C-terminus. Since decreasing the hydrophilic character of C-6, -7 and -8 of the C-ring of morphine leads to increased analgesic activity, the increased activity of the more hydrophobic chloromethyl ketones supports the suggestion that the Leu('5) carboxyl group of enkephalins overlaps the 6-hydroxyl group of morphine in the region of the C-ring. The biological activity of analogs substituted at the C-terminus with lysine, or a derivative of lysine containing either a formyl-, acetyl-, octanoyl-, or carbobenzyoxy-group attached to the (epsilon)-amino function, confirmed the importance of a hydrophobic group at that position. The most potent analog, D-Ala('2), N-(epsilon)-carbobenzyoxy-Lys('5)-enkephalin, was found to be almost three times more potent than the corresponding leucine enkephalin compound in the guinea pig ileum assay. These studies show that relatively large groups can be accommodated at the terminal position, without loss of activity. The synthesis of an LHRH precursor, in which the amino acid side chain functional groups are blocked by protecting groups, is also reported. S

Reprint of the 1965 ed. (3rd ed, with new preface