DANCE: a deep learning library and benchmark platform for single-cell analysis.

DANCE is the first standard, generic, and extensible benchmark platform for accessing and evaluating computational methods across the spectrum of benchmark datasets for numerous single-cell analysis tasks. Currently, DANCE supports 3 modules and 8 popular tasks with 32 state-of-art methods on 21 benchmark datasets. People can easily reproduce the results of supported algorithms across major benchmark datasets via minimal efforts, such as using only one command line. In addition, DANCE provides an ecosystem of deep learning architectures and tools for researchers to facilitate their own model development. DANCE is an open-source Python package that welcomes all kinds of contributions.

Single-Cell Multimodal Prediction via Transformers.

The recent development of multimodal single-cell technology has made the possibility of acquiring multiple omics data from individual cells, thereby enabling a deeper understanding of cellular states and dynamics. Nevertheless, the proliferation of multimodal single-cell data also introduces tremendous challenges in modeling the complex interactions among different modalities. The recently advanced methods focus on constructing static interaction graphs and applying graph neural networks (GNNs) to learn from multimodal data. However, such static graphs can be suboptimal as they do not take advantage of the downstream task information; meanwhile GNNs also have some inherent limitations when deeply stacking GNN layers. To tackle these issues, in this work, we investigate how to leverage transformers for multimodal single-cell data in an end-to-end manner while exploiting downstream task information. In particular, we propose a scMoFormer framework which can readily incorporate external domain knowledge and model the interactions within each modality and cross modalities. Extensive experiments demonstrate that scMoFormer achieves superior performance on various benchmark datasets. Remarkably, scMoFormer won a Kaggle silver medal with the rank of 24/1221 (Top 2%) without ensemble in a NeurIPS 2022 competition. Our implementation is publicly available at Github.

Identification of priority conservation areas based on ecosystem services and systematic conservation planning analysis.

In order to reverse the trend of ecological deterioration and resolve the conflict between ecological conservation and economic development, it is necessary to evaluate the trends of ecosystem services (ESs) and unravel the relationship between ESs and environmental drivers and identify the priority areas for ESs. In this research, we used the Integrated Valuation of Ecosystem Services and Tradeoffs (InVEST) model to quantify the variation of four important ESs (water purification, water yield, soil conservation, and habitat quality) in the Dongting Lake Basin from 2000 to 2015. During the past 15 years, water yield was declined by 3.38% and soil conservation was increased by 1.45%. Water quality purification was deteriorated with the rise in phosphorus export (5.32%) and nitrogen export (4.09%). Meanwhile, habitat quality was decreased by 3.27%. Trade-offs occurred primarily among water yield and other ESs. Social-ecological drivers importance analysis found that water yield was primarily influenced by precipitation and temperature. By contrast, water purification and habitat quality were more affected by the distribution of land use and land cover (LULC). Soil conservation was closely related to precipitation and geographical factor. Based on the distribution of ESs and the intensity of human activities, we delineated priority areas for each ESs using the systematic conservation planning tool (Marxan). LULC shifted most dramatically in water yield reserves (6.49%) with a large amount of lands conversed to cropland (4.4%) and build-up land (0.27%), which further increased the risk of water scarcity, while LULC changed less in other ESs priority areas due to human activities. Our study helps develop conservation strategies within specific area cost-effectively and provides scientific support for future conservation program of ESs formulation and adjustment.

Truly random number generation based on measurement of phase noise of a laser.

We present a simple approach to realize truly random number generator based on measuring the phase noise of a single-mode vertical cavity surface emitting laser. The true randomness of the quantum phase noise originates from the spontaneous emission of photons and the random bit generation rate is ultimately limited only by the laser linewidth. With the final bit generation rate of 20 Mbit/s, the truly random bit sequence guaranteed by the uncertainty principle of quantum mechanics passes the three standard randomness tests (ENT, Diehard, and NIST Statistical Test Suites). Moreover, a continuously generated random bit sequence, with length up to 14 Gbit, is verified by two additional criteria for its true randomness.