Search for: All records

Spatial transcriptomics (ST) has emerged as a powerful technology for bridging histology imaging with gene expression profiling. However, its application has been limited by low throughput and the need for specialized experimental facilities. Prior works sought to predict ST from whole-slide histology images to accelerate this process, but they suffer from two major limitations. First, they do not explicitly model cell-cell interaction as they factorize the joint distribution of whole-slide ST data and predict the gene expression of each spot independently. Second, their encoders struggle with memory constraints due to the large number of spots (often exceeding 10,000) in typical ST datasets. Herein, we propose STFlow, a flow matching generative model that considers cell-cell interaction by modeling the joint distribution of gene expression of an entire slide. It also employs an efficient slide-level encoder with local spatial attention, enabling whole-slide processing without excessive memory overhead. On the recently curated HEST-1k and STImage-1K4M benchmarks, STFlow substantially outperforms state-of-the-art baselines and achieves over 18% relative improvements over the pathology foundation models. 

We examine the arsenic distribution and its influence on dopant activation in poly-crystalline CdTe1−xSex solar cell absorbers prepared by vapor transport deposition followed by CdCl2 annealing. For as-deposited CdTe:As, local-electrode atom probe (LEAP) tomography reveals non-uniform distributions of arsenic clusters in the top “doped” layers. Following CdCl2 annealing, secondary ion mass spectrometry suggests that arsenic has diffused into the entire CdTe layer, while LEAP tomography reveals dissolution of the clusters, with nearly uniform distribution of arsenic atoms in CdTe. Since the arsenic fraction (fAs) is 1 × 1018 cm−3, but the hole density ranges from 7.5 to 9.5 × 1015 cm−3, we hypothesize that a large fraction of the arsenic has been incorporated into interstitial sites or cadmium substitutional sites, resulting in limited dopant activation. 

Nucleic acid-based drugs like aptamers have recently demonstrated great therapeutic potential. However, experimental platforms for aptamer screening are costly, and the scarcity of labeled data presents a challenge for supervised methods to learn protein-aptamer binding. To this end, we develop an unsupervised learning approach based on the predicted pairwise contact map between a protein and a nucleic acid and demonstrate its effectiveness in protein-aptamer binding prediction. Our model is based on FAFormer, a novel equivariant transformer architecture that seamlessly integrates frame averaging (FA) within each transformer block. This integration allows our model to infuse geometric information into node features while preserving the spatial semantics of coordinates, leading to greater expressive power than standard FA models. Our results show that FAFormer outperforms existing equivariant models in contact map prediction across three protein complex datasets, with over 10% relative improvement. Moreover, we curate five real-world protein-aptamer interaction datasets and show that the contact map predicted by FAFormer serves as a strong binding indicator for aptamer screening. 

Semiconductor quantum dots (QDs) are nanostructures that can enhance the performance of electronic devices due to their 3D quantization. Typically, heterovalent impurities, or dopants, are added to semiconducting QDs to provide extra electrons and improve conductivity. Since each QD is expected to contain a few dopants, the extra electrons and their parent dopants have been difficult to locate. In this work, we investigate the spatial distribution of the extra electrons and their parent donors in epitaxial InAs/GaAs QDs using local-electrode atom-probe tomography and self-consistent Schrödinger–Poisson simulations in the effective mass approximation. Although dopants are provided in both layers, the ionized donors primarily reside outside of the QDs, providing extra electrons that are contained within the QDs. Indeed, due to the quantum confinement-induced enhancement of the donor ionization energy within the QDs, a lower fraction of dopants within the QDs are ionized. These findings suggest a pathway toward the development of 3D modulation-doped nanostructures. 

We have investigated the origins of photoluminescence from quantum dot (QD) layers prepared by alternating depositions of sub-monolayers and a few monolayers of size-mismatched species, termed as sub-monolayer (SML) epitaxy, in comparison with their Stranski–Krastanov (SK) QD counterparts. Using measured nanostructure sizes and local In-compositions from local-electrode atom probe tomography as input into self-consistent Schrödinger–Poisson simulations, we compute the 3D confinement energies, probability densities, and photoluminescence (PL) spectra for both InAs/GaAs SML- and SK-QD layers. A comparison of the computed and measured PL spectra suggests one-dimensional electron confinement, with significant 3D hole localization in the SML-QD layers that contribute to their enhanced PL efficiency in comparison to their SK-QD counterparts. 

We have investigated the influence of non-stoichiometry and local atomic environments on carrier transport in GaAs(N)Bi alloy films using local-electrode atom probe tomography (LEAP) in conjunction with time-resolved terahertz photoconductivity measurements. The local concentrations of N, Bi, and excess As, as well as Bi pair correlations, are quantified using LEAP. Using time-resolved THz photoconductivity measurements, we show that carrier transport is primarily limited by excess As, with the highest carrier mobilities for layers with yBi > 0.035.