Xuedong Zheng

Identifying cancer driver genes is a prerequisite for developing targeted therapies. Although deep learning methods can integrate multi-omics data to predict driver genes, their results reflect statistical associations and cannot determine whether a gene's expression change is a cause or consequence of cancer. This study combines GLIMS prediction with bidirectional two-sample Mendelian randomization causal validation, and conducts systematic analysis in lung adenocarcinoma, glioblastoma, and breast cancer. GLIMS was first used to integrate four types of omics data for candidate driver gene prediction, and the results were intersected with differentially expressed genes consistently identified by limma, DESeq2, and edgeR, yielding 216 candidate genes. Using cis-eQTLs from the eQTLGen Consortium as instrumental variables and cancer GWAS data as outcomes, forward MR was performed to test causal effects, followed by reverse MR or Steiger directionality test to exclude reverse causation. Thirteen causal driver genes were finally identified: 4 in LUAD (LRRK2, SPP1, MARCO, CAV1), 2 in GBM (MAP1LC3A, SMG7), and 7 in BRCA (PARP1, GSN, RUNX2, PPARGC1A, PLK1, ITSN1, CBX4). Sensitivity analysis detected no horizontal pleiotropy. GO and KEGG enrichment showed that candidate genes were concentrated in cell cycle. Seven genes were significantly associated with patient prognosis, and immune infiltration analysis revealed distinct association patterns between driver genes and the immune microenvironment across different cancer types. The identified causal driver genes may serve as candidate targets for precision oncology.

As substances carrying genetic information, the properties of DNA molecules, such as natural programmability and the ability of information storage, provide a brand-new implementation method for constructing biocompatible logic circuits at the nanoscale. This paper proposes a cooperative hairpin DNA strand displacement mechanism, and based on this mechanism, two primitive DNA logic gates including OR gate and INHIBIT gate are constructed. In addition, an OR-INHIBIT DNA circuit are realized by juxtaposing these basic DNA logic gates in an isomorphic way. The transformation of the functions of basic logic gates and logic circuits is achieved by changing the auxiliary signal strands and input strands, which greatly reduces the complexity of DNA circuits. This construction method can enrich the toolbox for DNA circuit construction and contribute to the development of fields such as biosensors and biocomputing.

DNA computing has gained widespread attention for leveraging the unique properties of DNA molecules to perform computational operations. As a fundamental tool for analyzing data and optimizing models, matrix operation plays an important role in intensive computational tasks and is a focus of DNA-based numerical computation. However, complex computing tasks are often achieved through transmitting and processing signals successively, which requires matrix operations to perform calculations sequentially. Therefore, it is important to find a way to perform successive matrix operations to ensure computational sustainability in molecular computing. In this paper, we present a successive DNA matrix operations method based on the mechanism of combinatorial allosteric DNA strand displacement. In this mechanism, the input signal and the output signal are completely decoupled in the base arrangement of the DNA domain, which makes it easy to implement successive DNA matrix operations and easily realize the connection of DNA signal processing units. Based on this mechanism, some basic DNA logic gates, such as AND gate, OR gate, and INHIBIT gate, were constructed first, then Boolean matrix multiplication was realized and, finally, matrix-chain multiplication was completed to illustrate successive DNA matrix operations. This study provides a new way to implement successive DNA matrix operations and enriches the toolbox for achieving intensive computational tasks through molecular computing.

Structure-based molecular regulations have been widely adopted to modulate protein networks in cells and recently developed to control allosteric DNA operations in vitro. However, current examples of programmable allosteric signal transmission through integrated DNA networks are stringently constrained by specific design requirements. Developing a new, more general, and programmable scheme for establishing allosteric DNA networks remains challenging. Here, we developed a general strategy for programmable allosteric DNA regulations that can be finely tuned by varying the dimensions, positions, and number of conformational signals. By programming the allosteric signals, we realized fan-out/fan-in DNA gates and multiple-layer DNA cascading networks, as well as expanding the approach to long-range allosteric signal transmission through tunable DNA origami nanomachines ~100 nm in size. This strategy will enable programmable and complex allosteric DNA networks and nanodevices for nanoengineering, chemical, and biomedical applications displaying sense-compute-actuate molecular functionalities.

Preemptor blocks the strand displacement reaction by acting on DNA complex, not by directly hybridizing with the worker.

DNA computing plays an important role in nanotechnology due to the unique programmability and parallelism of DNA molecules. As an important tool to realize DNA computation, various logic computing devices have great application potential. The application of DNAzyme makes the achievements in the field of logical computing more diverse. In order to improve the efficiency of the logical units run by DNAzyme, we proposed a strategy to regulate the DNA circuit by the conformational change of the E6-type DNAzyme recognition arms driven by Mg²⁺. This strategy changes the single mode of DNAzyme signal transmission, extends the functions of E6-type DNAzyme, and saves the time of signal transmission in the molecular scale. To verify the feasibility of this strategy, first, we constructed DNA logic gates (YES, OR, and AND). Second, we cascade different logic gates (YES–YES, YES–AND) to prove the scalability. Finally, a self-catalytic DNA circuit is established. Through the experimental results, we verified that this DNAzyme regulation strategy relatively reduces the cost of logic circuits to some extent and significantly increases the reaction rate, and can also be used to indicate the range of Mg²⁺ concentrations. This research strategy provides new thinking for logical computing and explores new directions for detection and biosensors.