Parallel convolutional processing using an integrated photonic tensor core.

With the proliferation of ultrahigh-speed mobile networks and internet-connected devices, along with the rise of artificial intelligence (AI)1, the world is generating exponentially increasing amounts of data that need to be processed in a fast and efficient way. Highly parallelized, fast and scalable hardware is therefore becoming progressively more important2. Here we demonstrate a computationally specific integrated photonic hardware accelerator (tensor core) that is capable of operating at speeds of trillions of multiply-accumulate operations per second (1012 MAC operations per second or tera-MACs per second). The tensor core can be considered as the optical analogue of an application-specific integrated circuit (ASIC). It achieves parallelized photonic in-memory computing using phase-change-material memory arrays and photonic chip-based optical frequency combs (soliton microcombs3). The computation is reduced to measuring the optical transmission of reconfigurable and non-resonant passive components and can operate at a bandwidth exceeding 14 gigahertz, limited only by the speed of the modulators and photodetectors. Given recent advances in hybrid integration of soliton microcombs at microwave line rates3-5, ultralow-loss silicon nitride waveguides6,7, and high-speed on-chip detectors and modulators, our approach provides a path towards full complementary metal-oxide-semiconductor (CMOS) wafer-scale integration of the photonic tensor core. Although we focus on convolutional processing, more generally our results indicate the potential of integrated photonics for parallel, fast, and efficient computational hardware in data-heavy AI applications such as autonomous driving, live video processing, and next-generation cloud computing services.

Direct molecular versus culture-based assessment of Gram-positive cocci in biopsies of patients with major abscesses and diabetic foot infections.

Major abscesses and diabetic foot infections (DFIs) are predominant subtypes of complicated skin and skin structure infections (cSSSIs), and are mainly caused by Staphylococcus aureus and ß-hemolytic streptococci. This study evaluates the potential benefit of direct pathogen-specific real-time polymerase chain reaction (PCR) assays in the identification of causative organisms of cSSSIs. One-hundred and fifty major abscess and 128 DFI biopsy samples were collected and microbial DNA was extracted by using the Universal Microbe Detection kit for tissue samples. Pathogen-specific PCRs were developed for S. aureus and its virulence factor Panton-Valentine leukocidin (PVL), Streptococcus pyogenes, S. agalactiae, S. dysgalactiae, and the S. anginosus group. Identification by pathogen-specific PCRs was compared to routine culture and both methods were considered as the gold standard for determination of the sensitivity and specificity of each assay. Direct real-time PCR assays of biopsy samples resulted in a 34 % higher detection of S. aureus, 37 % higher detection of S. pyogenes, 18 % higher detection of S. agalactiae, 4 % higher detection of S. dysgalactiae subspecies equisimilis, and 7 % higher detection of the S. anginosus group, compared to routine bacterial culture. The presence of PVL was mainly confined to S. aureus isolated from major abscess but not DFI biopsy samples. In conclusion, our pathogen-specific real-time PCR assays had a higher yield than culture methods and could be an additional method for the detection of relevant causative pathogens in biopsies.

Polymorphisms in cytokine genes IL6, TNF, IL10, IL17A and IFNG influence susceptibility to complicated skin and skin structure infections.

Complicated skin and skin structure infections (cSSSIs) are caused by Gram-positive and Gram-negative, aerobic and anaerobic pathogens, with a polymicrobial aetiology being frequent. Recognition of invading pathogens by the immune system results in the production of pro- and anti-inflammatory cytokines, which are extremely important for intercellular communication and control of infection. This study assessed whether genetic variation in genes encoding cytokines influences the susceptibility to cSSSIs. For the association study, 318 patients with cSSSI and 328 healthy controls were genotyped for single nucleotide polymorphisms (SNPs) in cytokine genes IL1A, IL1B, IL1RN, TNF, IL10, IL17A, IL17F and IFNG. For immunological validation, peripheral blood mononuclear cells (PBMCs) from 74 healthy individuals, genotyped for SNPs of interest, were stimulated with Staphylococcus aureus or Escherichia coli and corresponding cytokine levels were determined by enzyme-linked immunosorbent assay (ELISA). Polymorphisms IL6 rs1800797, TNF rs1800629, IL10 rs1800871, IL17A rs8193036 and IFNG rs2069705 influenced susceptibility to cSSSIs. No differences in cytokine responses, stratified for genotype, were detected after PBMC stimulation. No association with cSSSIs was observed for polymorphisms IL1A rs17561 and rs1800587, IL1B rs16944 and rs1143627, IL1RN rs4251961, TNF rs361525, IL10 rs1800896, IL17A rs2275913 and IL17F rs763780. In conclusion, polymorphisms in IL6, TNF, IL10, IL17A and IFNG are associated with susceptibility to cSSSIs.

Different patterns of Toll-like receptor 2 polymorphisms in populations of various ethnic and geographic origins.

Upon the invasion of the host by microorganisms, innate immunity is triggered through pathogen recognition by pattern recognition receptors (PRRs). Toll-like receptors (TLRs) are the best-studied class of PRRs, and they recognize specific pathogen-associated molecular patterns (PAMPs) from various microorganisms. A large number of studies have shown that genetic variation in TLRs may influence susceptibility to infections. We assessed the genetic variation of TLR2, which encodes one of the most important TLRs, in various populations around the globe and correlated it with changes in the function of the molecule. The three best-known nonsynonymous TLR2 polymorphisms (1892C>A, 2029C>T, and 2258G>A) were assessed in different populations from the main continental masses: Romanians, Vlax-Roma, Dutch (European populations), Han Chinese (East Asia), Dogon, Fulani (Africa), and Trio Indians (America). The 2029C>T polymorphism was absent in both European and non-European populations, with the exception of the Vlax-Roma, suggesting that this polymorphism most likely arose in Indo-Aryan people after migration into South Asia. The 1892C>A polymorphism that was found exclusively in European populations, but not in Asian, African, or American volunteers, probably occurred in proto-Indo-Europeans. Interestingly, 2258G>A was present only in Europeans, including Vlax-Roma, but at a very low frequency. The differential pattern of the TLR2 polymorphisms in various populations may explain some of the differences in susceptibility to infections between these populations.