Search

Despite the striking success of general protein folding models such as AlphaFold2(AF2, Jumper et al. (2021)), the accurate computational modeling of antibody-antigen complexes remains a challenging task. In this paper, we first analyze AF2's primary loss function, known as the Frame Aligned Point Error (FAPE), and raise a previously overlooked issue that FAPE tends to face gradient vanishing problem on high-rotational-error targets. To address this fundamental limitation, we propose a novel geodesic loss called Frame Aligned Frame Error (FAFE, denoted as F2E to distinguish from FAPE), which enables the model to better optimize both the rotational and translational errors between two frames. We then prove that F2E can be reformulated as a group-aware geodesic loss, which translates the optimization of the residue-to-residue error to optimizing group-to-group geodesic frame distance. By fine-tuning AF2 with our proposed new loss function, we attain a correct rate of 52.3\% (DockQ $>$ 0.23) on an evaluation set and 43.8\% correct rate on a subset with low homology, with substantial improvement over AF2 by 182\% and 100\% respectively.

Peptides, short chains of amino acid residues, play a vital role in numerous biological processes by interacting with other target molecules, offering substantial potential in drug discovery. In this work, we present PepFlow, the first multi-modal deep generative model grounded in the flow-matching framework for the design of full-atom peptides that target specific protein receptors. Drawing inspiration from the crucial roles of residue backbone orientations and side-chain dynamics in protein-peptide interactions, we characterize the peptide structure using rigid backbone frames within the $\mathrm{SE}(3)$ manifold and side-chain angles on high-dimensional tori. Furthermore, we represent discrete residue types in the peptide sequence as categorical distributions on the probability simplex. By learning the joint distributions of each modality using derived flows and vector fields on corresponding manifolds, our method excels in the fine-grained design of full-atom peptides. Harnessing the multi-modal paradigm, our approach adeptly tackles various tasks such as fix-backbone sequence design and side-chain packing through partial sampling. Through meticulously crafted experiments, we demonstrate that PepFlow exhibits superior performance in comprehensive benchmarks, highlighting its significant potential in computational peptide design and analysis., Comment: ICML 2024

We introduce Statistical Flow Matching (SFM), a novel and mathematically rigorous flow-matching framework on the manifold of parameterized probability measures inspired by the results from information geometry. We demonstrate the effectiveness of our method on the discrete generation problem by instantiating SFM on the manifold of categorical distributions whose geometric properties remain unexplored in previous discrete generative models. Utilizing the Fisher information metric, we equip the manifold with a Riemannian structure whose intrinsic geometries are effectively leveraged by following the shortest paths of geodesics. We develop an efficient training and sampling algorithm that overcomes numerical stability issues with a diffeomorphism between manifolds. Our distinctive geometric perspective of statistical manifolds allows us to apply optimal transport during training and interpret SFM as following the steepest direction of the natural gradient. Unlike previous models that rely on variational bounds for likelihood estimation, SFM enjoys the exact likelihood calculation for arbitrary probability measures. We manifest that SFM can learn more complex patterns on the statistical manifold where existing models often fail due to strong prior assumptions. Comprehensive experiments on real-world generative tasks ranging from image, text to biological domains further demonstrate that SFM achieves higher sampling quality and likelihood than other discrete diffusion or flow-based models., Comment: Accepted to NeurIPS 2024 as a conference paper

Electrical readout of magnetic states is a key to realize novel spintronics devices for efficient computing and data storage. Unidirectional magnetoresistance (UMR) in bilayer systems, consisting of a spin source material and a magnetic layer, refers to a change in the longitudinal resistance upon the reversal of magnetization, which typically originates from the interaction of spin-current and magnetization at the interface. Because of UMR s linear dependence on applied charge current and magnetization, it can be used to electrically read the magnetization state. However, in conventional spin source materials, the spin polarization of an electric field induced spin current is restricted to be in the film plane and hence the ensuing UMR can only respond to the in plane component of the magnetization. On the other hand, magnets with perpendicular magnetic anisotropy (PMA) are highly desired for magnetic memory and spin-logic devices, while the electrical read out of PMA magnets through UMR is critically missing. Here, we report the discovery of an unconventional UMR in bilayer heterostructures of a topological semimetal (WTe2) and a PMA ferromagnetic insulator (Cr2Ge2Te6, CGT), which allows to electrically read the up and down magnetic states of the CGT layer by measuring the longitudinal resistance. Our theoretical calculations based on a tight binding model show that the unconventional UMR originates from the interplay of crystal symmetry breaking in WTe2 and magnetic exchange interaction across the WTe2 and CGT interface. Combining with the ability of WTe2 to obtain magnetic field free switching of the PMA magnets, our discoveries open an exciting pathway to achieve two terminal magnetic memory devices that operate solely on the spin orbit torque and UMR, which is critical for developing next-generation non volatile and low power consumption data storage technologies.

The efficiency of phonon-mediated heat transport is limited by the intrinsic atomistic properties of materials, seemingly providing an upper limit to heat transfer in materials and across their interfaces. The typical speeds of conductive transport, which are inherently limited by the chemical bonds and atomic masses, dictate how quickly heat will move in solids. Given that phonon-polaritons, or coupled phonon-photon modes, can propagate at speeds approaching 1 percent of the speed of light - orders of magnitude faster than transport within a pure diffusive phonon conductor - we demonstrate that volume-confined, hyperbolic phonon-polariton(HPhP) modes supported by many biaxial polar crystals can couple energy across solid-solid interfaces at an order of magnitude higher rates than phonon-phonon conduction alone. Using pump-probe thermoreflectance with a mid-infrared, tunable, probe pulse with sub-picosecond resolution, we demonstrate remote and spectrally selective excitation of the HPhP modes in hexagonal boron nitride in response to radiative heating from a thermally emitting gold source. Our work demonstrates a new avenue for interfacial heat transfer based on broadband radiative coupling from a hot spot in a gold film to hBN HPhPs, independent of the broad spectral mismatch between the pump(visible) and probe(mid-IR) pulses employed. This methodology can be used to bypass the intrinsically limiting phonon-phonon conductive pathway, thus providing an alternative means of heat transfer across interfaces. Further, our time-resolved measurements of the temperature changes of the HPhP modes in hBN show that through polaritonic coupling, a material can transfer heat across and away from an interface at rates orders of magnitude faster than diffusive phonon speeds intrinsic to the material, thus demonstrating a pronounced thermal transport enhancement in hBN via phonon-polariton coupling.

End-to-end autonomous driving recently emerged as a promising research direction to target autonomy from a full-stack perspective. Along this line, many of the latest works follow an open-loop evaluation setting on nuScenes to study the planning behavior. In this paper, we delve deeper into the problem by conducting thorough analyses and demystifying more devils in the details. We initially observed that the nuScenes dataset, characterized by relatively simple driving scenarios, leads to an under-utilization of perception information in end-to-end models incorporating ego status, such as the ego vehicle's velocity. These models tend to rely predominantly on the ego vehicle's status for future path planning. Beyond the limitations of the dataset, we also note that current metrics do not comprehensively assess the planning quality, leading to potentially biased conclusions drawn from existing benchmarks. To address this issue, we introduce a new metric to evaluate whether the predicted trajectories adhere to the road. We further propose a simple baseline able to achieve competitive results without relying on perception annotations. Given the current limitations on the benchmark and metrics, we suggest the community reassess relevant prevailing research and be cautious whether the continued pursuit of state-of-the-art would yield convincing and universal conclusions. Code and models are available at \url{https://github.com/NVlabs/BEV-Planner}, Comment: Accept to cvpr 2024

In this work, we report on the growth of hexagonal boron nitride (hBN) crystals from an iron flux at atmospheric pressure and high temperature and demonstrate that (i) the entire sheet of hBN crystals can be detached from the metal in a single step using hydrochloric acid and that (ii) these hBN crystals allow the fabrication of high carrier mobility graphene devices. By combining spatially-resolved confocal Raman spectroscopy and electrical transport measurements, we confirm the excellent quality of these crystals for high-performance hBN-graphene-based van der Waals heterostructures. The full width at half maximum of the graphene Raman 2D peak is as low as 16 cm$^{-1}$, and the room temperature charge carrier mobilitiy is around 80000 cm$^2$/(Vs) at a carrier density 1$\times$10$^{12}$cm$^{-12}$. This is fully comparable with devices of similar dimensions fabricated using crystalline hBN synthesized by the high pressure and high temperature method. Finally, we show that for exfoliated high-quality hBN flakes with a thickness between 20 nm and 40 nm the line width of the hBN Raman peak, in contrast to the graphene 2D line width, is not useful for benchmarking hBN in high mobility graphene devices., Comment: 7 pages, 4 figures

Unsupervised domain adaptation person re-identification (Re-ID) aims to identify pedestrian images within an unlabeled target domain with an auxiliary labeled source-domain dataset. Many existing works attempt to recover reliable identity information by considering multiple homogeneous networks. And take these generated labels to train the model in the target domain. However, these homogeneous networks identify people in approximate subspaces and equally exchange their knowledge with others or their mean net to improve their ability, inevitably limiting the scope of available knowledge and putting them into the same mistake. This paper proposes a Dual-level Asymmetric Mutual Learning method (DAML) to learn discriminative representations from a broader knowledge scope with diverse embedding spaces. Specifically, two heterogeneous networks mutually learn knowledge from asymmetric subspaces through the pseudo label generation in a hard distillation manner. The knowledge transfer between two networks is based on an asymmetric mutual learning manner. The teacher network learns to identify both the target and source domain while adapting to the target domain distribution based on the knowledge of the student. Meanwhile, the student network is trained on the target dataset and employs the ground-truth label through the knowledge of the teacher. Extensive experiments in Market-1501, CUHK-SYSU, and MSMT17 public datasets verified the superiority of DAML over state-of-the-arts.

Equivariance has been a long-standing concern in various fields ranging from computer vision to physical modeling. Most previous methods struggle with generality, simplicity, and expressiveness -- some are designed ad hoc for specific data types, some are too complex to be accessible, and some sacrifice flexible transformations. In this work, we propose a novel and simple framework to achieve equivariance for point cloud analysis based on the message passing (graph neural network) scheme. We find the equivariant property could be obtained by introducing an orientation for each point to decouple the relative position for each point from the global pose of the entire point cloud. Therefore, we extend current message passing networks with a module that learns orientations for each point. Before aggregating information from the neighbors of a point, the networks transforms the neighbors' coordinates based on the point's learned orientations. We provide formal proofs to show the equivariance of the proposed framework. Empirically, we demonstrate that our proposed method is competitive on both point cloud analysis and physical modeling tasks. Code is available at https://github.com/luost26/Equivariant-OrientedMP ., Comment: Accepted to CVPR 2022

A protein performs biological functions by folding to a particular 3D structure. To accurately model the protein structures, both the overall geometric topology and local fine-grained relations between amino acids (e.g. side-chain torsion angles and inter-amino-acid orientations) should be carefully considered. In this work, we propose the Directed Weight Neural Network for better capturing geometric relations among different amino acids. Extending a single weight from a scalar to a 3D directed vector, our new framework supports a rich set of geometric operations on both classical and SO(3)--representation features, on top of which we construct a perceptron unit for processing amino-acid information. In addition, we introduce an equivariant message passing paradigm on proteins for plugging the directed weight perceptrons into existing Graph Neural Networks, showing superior versatility in maintaining SO(3)-equivariance at the global scale. Experiments show that our network has remarkably better expressiveness in representing geometric relations in comparison to classical neural networks and the (globally) equivariant networks. It also achieves state-of-the-art performance on various computational biology applications related to protein 3D structures.