Transfer Method Research Articles

Reference-free structural variant detection in microbiomes via long-read co-assembly graphs.

The study of bacterial genome dynamics is vital for understanding the mechanisms underlying microbial adaptation, growth, and their impact on host phenotype. Structural variants (SVs), genomic alterations of 50 base pairs or more, play a pivotal role in driving evolutionary processes and maintaining genomic heterogeneity within bacterial populations. While SV detection in isolate genomes is relatively straightforward, metagenomes present broader challenges due to the absence of clear reference genomes and the presence of mixed strains. In response, our proposed method rhea, forgoes reference genomes and metagenome-assembled genomes (MAGs) by encompassing all metagenomic samples in a series (time or other metric) into a single co-assembly graph. The log fold change in graph coverage between successive samples is then calculated to call SVs that are thriving or declining. We show rhea to outperform existing methods for SV and horizontal gene transfer (HGT) detection in two simulated mock metagenomes, particularly as the simulated reads diverge from reference genomes and an increase in strain diversity is incorporated. We additionally demonstrate use cases for rhea on series metagenomic data of environmental and fermented food microbiomes to detect specific sequence alterations between successive time and temperature samples, suggesting host advantage. Our approach leverages previous work in assembly graph structural and coverage patterns to provide versatility in studying SVs across diverse and poorly characterized microbial communities for more comprehensive insights into microbial gene flux. rhea is open source and available at: https://github.com/treangenlab/rhea.

Prostate cancer diagnosis based on multi-parametric MRI, clinical and pathological factors using deep learning

Prostate cancer is one of the most common and fatal diseases among men, and its early diagnosis can have a significant impact on the treatment process and prevent mortality. Since it does not have apparent clinical symptoms in the early stages, it is difficult to diagnose. In addition, the disagreement of experts in the analysis of magnetic resonance images is also a significant challenge. In recent years, various research has shown that deep learning, especially convolutional neural networks, has appeared successfully in machine vision (especially in medical image analysis). In this research, a deep learning approach was used on multi-parameter magnetic resonance images, and the synergistic effect of clinical and pathological data on the accuracy of the model was investigated. The data were collected from Trita Hospital in Tehran, which included 343 patients (data augmentation and learning transfer methods were used during the process). In the designed model, four different types of images are analyzed with four separate ResNet50 deep convolutional networks, and their extracted features are transferred to a fully connected neural network and combined with clinical and pathological features. In the model without clinical and pathological data, the maximum accuracy reached 88%, but by adding these data, the accuracy increased to 96%, which shows the significant impact of clinical and pathological data on the accuracy of diagnosis.

Approach to Detect Windows Malware Based on Malicious Tendency Image and ResNet Algorithm

Timely detection of self-replicating malware in the high market share Windows operating system can effectively prevent personal or corporate financial losses. The form and characteristics of malware are constantly evolving, leading to a concept drift issue that gradually decreases the effectiveness of traditional detection methods. Therefore, we propose WinMDet, a Windows malware detection method based on malicious tendency image and ResNet algorithm. First, to tackle the complexity and difficulty in accurately characterizing malware features, WinMDet retains detailed malware features and encodes them into malicious tendency images to better describe malware across different periods. Secondly, WinMDet utilizes previously generated malicious tendency images to train the initial detection model. Then, to alleviate the issue of malware concept drift, WinMDet employs Local Maximum Mean Discrepancy (LMMD) as the criterion for model transfer, enhancing the initial detection model’s ability to distinguish between malware and benign software. We conducted a comprehensive evaluation of WinMDet using common metrics such as accuracy, precision and recall. The results indicate that WinMDet performs remarkably well in terms of accuracy, exceeding 82%. Additionally, significant improvements were observed in precision and recall, surpassing 82.42% and 82.06%, respectively. After employing our LMMD-based transfer method, the initial detection model improved the detection accuracy of malware in 2021 and 2022 by approximately 4.22% to 8.06%. The false negative rate decreased by at most 4.34%, and the false positive rate decreased by at most 4.61%.

CRISPR-mediated editing of β-lactoglobulin (BLG) gene in buffalo

Milk is a good source of nutrition but is also a source of allergenic proteins such as α-lactalbumin, β-lactoglobulin (BLG), casein, and immunoglobulins. The Clustered Regularly Interspaced Short Palindromic Repeats (CRISPR)/Cas technology has the potential to edit any gene, including milk allergens. Previously, CRISPR/Cas has been successfully employed in dairy cows and goats, but buffaloes remain unexplored for any milk trait. In this study, we utilized the CRISPR/Cas9 system to edit the major milk allergen BLG gene in buffaloes. First, the editing efficiency of designed sgRNAs was tested in fibroblast cells using the T7E assay and Sanger sequencing. The most effective sgRNA was selected to generate clonal lines of BLG-edited cells. Analysis of 15 single-cell clones, through TA cloning and Sanger sequencing, revealed that 7 clones exhibited bi-allelic (−/−) heterozygous, bi-allelic (−/−) homozygous, and mono-allelic (−/+) disruptions in BLG. Bioinformatics prediction analysis confirmed that non-multiple-of-3 edited nucleotide cell clones have frame shifts and early truncation of BLG protein, while multiple-of-3 edited nucleotides resulted in slightly disoriented protein structures. Somatic cell nuclear transfer (SCNT) method was used to produce blastocyst-stage embryos that have similar developmental rates and quality with wild-type embryos. This study demonstrated the successful bi-allelic editing (−/−) of BLG in buffalo cells through CRISPR/Cas, followed by the production of BLG-edited blastocyst stage embryos using SCNT. With CRISPR and SCNT methods described herein, our long-term goal is to generate gene-edited buffaloes with BLG-free milk.

Sequence-specific dynamic DNA bending explains mitochondrial TFAM’s dual role in DNA packaging and transcription initiation

Mitochondrial transcription factor A (TFAM) employs DNA bending to package mitochondrial DNA (mtDNA) into nucleoids and recruit mitochondrial RNA polymerase (POLRMT) at specific promoter sites, light strand promoter (LSP) and heavy strand promoter (HSP). Herein, we characterize the conformational dynamics of TFAM on promoter and non-promoter sequences using single-molecule fluorescence resonance energy transfer (smFRET) and single-molecule protein-induced fluorescence enhancement (smPIFE) methods. The DNA-TFAM complexes dynamically transition between partially and fully bent DNA conformational states. The bending/unbending transition rates and bending stability are DNA sequence-dependent—LSP forms the most stable fully bent complex and the non-specific sequence the least, which correlates with the lifetimes and affinities of TFAM with these DNA sequences. By quantifying the dynamic nature of the DNA-TFAM complexes, our study provides insights into how TFAM acts as a multifunctional protein through the DNA bending states to achieve sequence specificity and fidelity in mitochondrial transcription while performing mtDNA packaging.

Local context attention learning for fine-grained scene graph generation

Fine-grained scene graph generation aims to parse the objects and their fine-grained relationships within scenes. Despite the significant progress in recent years, their performance is still limited by two major issues: (1) ambiguous perception under a global view; (2) the lack of reliable, fine-grained annotations. We argue that understanding the local context is important in addressing the two issues. However, previous works often overlook it, which limits their effectiveness in fine-grained scene graph generation. To tackle this challenge, we introduce a Local-context Attention Learning method that concentrates on local context and can generate high-reliability, fine-grained annotations. It comprises two components: (1) The Fine-grained Location Attention Network (FLAN), a multi-branch network that encompasses global and local branches, can attend to local informative context and perceive granularity levels in different regions, thereby adaptively enhancing the learning of fine-grained locations. (2) The Fine-grained Location Label Transfer (FLLT) method identifies coarse-grained labels inconsistent with the local context and determines which labels should be transferred through the global confidence thresholding strategy, finally transferring them to reliable local context-consistent fine-grained ones. Experiments conducted on the Visual Genome, OpenImage, and GQA-200 datasets show that the proposed methods achieve significant improvements on the fine-grained scene graph generation task. By addressing the challenge mentioned above, our method also achieves state-of-the-art performances on the three datasets.

Ice-Enabled Transfer of Graphene on Copper Substrates Enhanced by Electric Field and Cu2O.

Graphene films grown by the chemical vapor deposition (CVD) method suffer from contamination and damage during transfer. Herein, an innovative ice-enabled transfer method under an applied electric field and in the presence of Cu2O (or Cu2O-Electric-field Ice Transfer, abbreviated as CEIT) is developed. Ice serves as a pollution-free transfer medium while water molecules under the electric field fully wet the graphene surface for a bolstered adhesion force between the ice and graphene. Cu2O is used to reduce the adhesion force between graphene and copper. The combined methodology in CEIT ensures complete separation and clean transfer of graphene, resulting in successfully transferred graphene to various substrates, including polydimethylsiloxane (PDMS), Teflon, and C4F8 without pollution. The graphene obtained via CEIT is utilized to fabricate field-effect transistors with electrical performances comparable to that of intrinsic graphene characterized by small Dirac points and high carrier mobility. The carrier mobility of the transferred graphene reaches 9090 cm2V-1s-1, demonstrating a superior carrier mobility over that from other dry transfer methods. In a nutshell, the proposed clean and efficient transfer method holds great potential for future applications of graphene.

Selective policy transfer in multi-agent systems with sparse interactions

Previously trained single-agent strategies, which are considerably easier to acquire than multi-agent strategies, are instructive for multi-agent reinforcement learning, especially when the interactions between agents are sparse. Traditional methods of knowledge transfer, from single-agent source tasks to multi-agent target tasks, typically rely on a pre-designed Markov-decision-process similarity function or metric function for evaluating the task similarity. In this study, we propose a selective policy transfer (SPOT) algorithm that eliminates the need for the manual crafting of a metric function to assess the similarity between source and target tasks. The SPOT algorithm enables agents to autonomously determine when and which policy to transfer using well-trained single-agent policies as options in the training process. We introduced a multi-agent policy-learning option in the option library, thus allowing the SPOT algorithm to leverage the transferred knowledge while concurrently learning new policies. The SPOT algorithm efficiently transfers sequential strategies in a few steps, thereby capturing high-level semantics. Experimental results obtained in both multi-agent arcade and multi-agent particle environments have demonstrated that the proposed algorithm outperforms the state-of-the-art methods in terms of jump-start and convergence speeds. Our evaluation indicators included the jump start, steps to threshold, cumulative reward, and asymptotic performance. Furthermore, visualizations of the strategies and termination functions of the agents aid in elucidating the operational principles of the SPOT algorithm.

Evolution and Driving Forces of Ecological Service Value in Response to Land Use Change in Tarim Basin, Northwest China

The main objective of protecting ecosystems and enhancing the supply of ecosystem services (ESs) is to quantify the value of ecological services. This article calculates the ecological service value (ESV) of the Tarim Basin over the past 40 years using the improved benefits transfer method of satellite remote sensing data, such as Landsat, analyzes the spatiotemporal evolution characteristics of ESV, and studies the driving mechanism of ESV changes using GeoDetector. Finally, the FLUS model was selected to predict the ecosystem service value until 2030, setting up three scenarios: the Baseline Scenario (BLS), the Cultivated Land Protection Scenario (CPS), and the Ecological Protection Scenario (EPS). The results indicate that (1) the ESV in the Tarim Basin decreased by USD 1248.21 million (−2.29%) from 1980 to 2020. The top three contributors are water bodies, wetlands, and grassland. (2) Waste treatment and water supply functions had the highest service value, accounting for 44.53% of the total contribution. The rank order of ecosystem functions in terms of their contribution to the total value of ESV was as follows, refining from high to low importance: water supply, waste treatment, biodiversity protection, climate regulation, soil formation, recreation and culture, gas regulation, food production, raw material. (3) The spatial differentiation driving factors of ESV were detected, with the following Q-values in descending order: net primary productivity (NPP) > normalized difference vegetation index (NDVI) > precipitation > aspect > temperature > slope > soil erosion > GDP > land use intensity > per capita GDP > population > human activity index. (4) The ESVs simulated under the three scenarios (BLS, CPS, and EPS) for 2030 were USD 51,133.9 million, USD 53,624.99 million, and USD 54,561.26 million, respectively. Compared with 2020, the ESVs of the three scenarios decreased as follows: BLS (USD 4209.33 million), CPS (USD 1718.24 million), and EPS USD (−781.97 million). These findings are significant for maintaining the integrity and sustainability of the large-scale ecosystem, where socioeconomic development and the fragile features of the natural ecosystem interact. Additionally, the study results provide a crucial foundation for governmental decision-makers, local residents, and environmental researchers in northwest China to promote sustainable development.

Benefits of Neurosurgical Teleconsults in the Management of Intracerebral Hemorrhage: Transfers and Transportation Cost Reduction

Our study explores the efficacy and economic benefits of neurosurgical teleconsultations in managing intracerebral hemorrhage (ICH), focusing on reducing unnecessary patient transfers and associated costs. We conducted a cost-savings analysis at our institution of a previously published pilot study involving a cohort of patients with ICH who were potential candidates for airlift to our tertiary care center but instead received neurosurgical consultation via teleconsultation to avoid the transfer. Data on patient demographics, distances, and costs were collected and analyzed to assess the economic impact of teleconsultations. The cohort comprised 14 patients; we noted significant cost savings from avoiding interhospital transfers, ranging from $84,346.52 to $120,495.03 per patient. Teleconsultations facilitated immediate, collaborative decision-making between healthcare providers at community hospitals and a tertiary care center, reducing the need for expensive air transportation and unnecessary hospital transfers. Neurosurgical teleconsultations offer a cost-effective alternative to traditional patient transfer methods for ICH management, providing substantial economic benefits while maintaining high physician and patient-family satisfaction levels. This study underscores the potential of our teleneurosurgery program to significantly reduce costs by reducing unnecessary financial burdens on patients' families and healthcare systems.

Genetic manipulations of nonmodel gut microbes.

Hundreds of microbiota gene expressions are significantly different between healthy and diseased humans. The "bottleneck" preventing a mechanistic dissection of how they affect host biology/disease is that many genes are encoded by nonmodel gut commensals and not genetically manipulatable. Approaches to efficiently identify their gene transfer methodologies and build their gene manipulation tools would enable mechanistic dissections of their impact on host physiology. This paper will introduce a step-by-step protocol to identify gene transfer conditions and build the gene manipulation tools for nonmodel gut microbes, focusing on Gram-negative Bacteroidia and Gram-positive Clostridia organisms. This protocol enables us to identify gene transfer methods and develop gene manipulation tools without prior knowledge of their genome sequences, by targeting bacterial 16s ribosomal RNAs or expanding their compatible replication origins combined with clustered regularly interspaced short palindromic repeats machinery. Such an efficient and generalizable approach will facilitate functional studies that causally connect gut microbiota genes to host diseases.

Theoretical analysis on the behavior of SPM anchor piles embedded in sand under oblique pullout load with emphasis on the padeye depth

Single-point mooring (SPM) system has been used in offshore engineering for its superiority in overcoming severe sea conditions. However, researches focused on the interaction mechanism between the anchor pile and soil in SPM systems remains insufficient, especially regarding the calculation method for oblique pullout bearing capacity that considers padeye depth. Based on the p-y curve method and load transfer method, this paper presents a simple prediction method for the behavior of anchor piles embedded in sand under oblique pullout load. The theoretical predictions are validated through comparing with results from the three-dimensional finite element method and existing centrifuge model tests. The effect of the loading angle and the padeye depth are analyzed and discussed. A negative bending moment peak and axial force discontinuity phenomenon appear at the padeye depth. The optimal padeye depth for anchor piles in this study ranges from approximately 0.53 to 0.80 times the pile length. When the padeye depth is less than 0.33 times of the pile length, the ultimate bearing capacity of the anchor pile increases with increasing load angles. Conversely, when the padeye depth exceeds 0.33 times the pile length, the ultimate bearing capacity decreases with the increasing load angles.

Two-Dimensional GeS/SnSe2 Tunneling Photodiode with Bidirectional Photoresponse and High Polarization Sensitivity.

A two-dimensional (2D) broken-gap (type-III) p-n heterojunction has a unique charge transport mechanism because of nonoverlapping energy bands. In light of this, type-III band alignment can be used in tunneling field-effect transistors (TFETs) and Esaki diodes with tunable operation and low consumption by highlighting the advantages of tunneling mechanisms. In recent years, 2D tunneling photodiodes have gradually attracted attention for novel optoelectronic performance with a combination of strong light-matter interaction and tunable band alignment. However, an in-depth understanding of the tunneling mechanisms should be further investigated, especially for developing electronic and optoelectronic applications. Here, we report a type-III tunneling photodiode based on a 2D multilayered p-GeS/n+-SnSe2 heterostructure, which is first fabricated by the mechanical exfoliation and dry transfer method. Through the Simmons approximation, its various tunneling transport mechanisms dependent on bias and light are demonstrated as the origin of excellent bidirectional photoresponse performance. Moreover, compared to the traditional p-n photodiode, the device enables bidirectional photoresponse capability, including maximum responsivity values of 43 and 8.7 A/W at Vds = 1 and -1 V, respectively, with distinctive photoactive regions from the scanning photocurrent mapping. Noticeably, benefiting from the in-plane anisotropic structure of GeS, the device exhibits an enhanced photocurrent anisotropic ratio of 9, driven by the broader depletion region at Vds = -3 V under 635 nm irradiation. Above all, the results suggest that our designed architecture can be potentially applied to CMOS imaging sensors and polarization-sensitive photodetectors.

A Denoising Convolutional Autoencoder for SNR Enhancement in Chemical Exchange Saturation Transfer imaging: (DCAE-CEST).

To develop a SNR enhancement method for chemical exchange saturation transfer (CEST) imaging using a denoising convolutional autoencoder (DCAE), and compare its performance with state-of-the-art denoising methods. The DCAE-CEST model encompasses an encoder and a decoder network. The encoder learns features from the input CEST Z-spectrum via a series of 1D convolutions, nonlinearity applications and pooling. Subsequently, the decoder reconstructs an output denoised Z-spectrum using a series of up-sampling and convolution layers. The DCAE-CEST model underwent multistage training in an environment constrained by Kullback-Leibler divergence, while ensuring data adaptability through context learning using Principal Component Analysis processed Z-spectrum as a reference. The model was trained using simulated Z-spectra, and its performance was evaluated using both simulated data and in-vivo data from an animal tumor model. Maps of amide proton transfer (APT) and nuclear Overhauser enhancement (NOE) effects were quantified using the multiple-pool Lorentzian fit, along with an apparent exchange-dependent relaxation metric. In digital phantom experiments, the DCAE-CEST method exhibited superior performance, surpassing existing denoising techniques, as indicated by the peak SNR and Structural Similarity Index. Additionally, in vivo data further confirms the effectiveness of the DCAE-CEST in denoising the APT and NOE maps when compared to other methods. While no significant difference was observed in APT between tumors and normal tissues, there was a significant difference in NOE, consistent with previous findings. The DCAE-CEST can learn the most important features of the CEST Z-spectrum and provide the most effective denoising solution compared to other methods.

Damage-free dry transfer method using stress engineering for high-performance flexible two- and three-dimensional electronics.

Advanced transfer printing technologies have enabled the fabrication of high-performance flexible and stretchable devices, revolutionizing many research fields including soft electronics, optoelectronics, bioelectronics and energy devices. Despite previous innovations, challenges remain, such as safety concerns due to toxic chemicals, the expensive equipment, film damage during the transfer process and difficulty in high-temperature processing. Thus a new transfer printing process is needed for the commercialization of high-performance soft electronic devices. Here we propose a damage-free dry transfer printing strategy based on stress control of the deposited thin films. First, stress-controlled metal bilayer films are deposited using direct current magnetron sputtering. Subsequently, mechanical bending is applied to facilitate the release of the metal bilayer by increasing the overall stress. Experimental and simulation studies elucidate the stress evolution mechanisms during the processes. By using this method, we successfully transfer metal thin films and high-temperature-treated oxide thin films onto flexible or stretchable substrates, enabling the fabrication of two-dimensional flexible electronic devices and three-dimensional multifunctional devices.

Efficient high-throughput method utilizing neural network potentials to calculate interaction energies, validated by clean transfer experiment of CVD graphene with polymer mixtures

In this study, we designed a computational model involving energy decomposition using ground state energy minimized geometries resulting from a general-purpose neural network potential (ANI-1ccx). The numerical simulations show a distribution of energies, which indicate a two-fold reduction in interaction energy and polarity shift in electrostatic interaction, highlighting the computational novelty in exploring over a million metastable configurations. Experimentally, we validate our model by observing that using a mixture of two distinct polymers in the wet transfer process reduces transfer-induced doping and strain on transferred CVD graphene compared to the conventional single polymer wet transfer method, primarily due to decreased polymer contamination from the transfer process. This reduction is linked to the decreased interaction energy in the mixture of polymethyl methacrylate and angelica lactone polymer on graphene.

Study of high-precision time transfer method enhanced by PPP-AR/PPP-RTK

Abstract With the continuous development and improvement of Global Navigation Satellite Systems (GNSS), the use of GNSS for high-precision time transfer has become increasingly mature. To delve into the contribution of information-enhanced GNSS PPP to time transfer performance, this paper focuses on the comprehensive evaluation and analysis of the time transfer performance of PPP-AR and PPP-RTK. Using GPS as a case study, experimental results demonstrate that the average success fixing rate of PPP integer ambiguity resolution across five stations is 93.57%, with initial fixing times consistently below 20 minutes. The stability of the station clock offset sequence of PPP-AR is better than that of PPP floating solution. Compared with the PPP floating solution, the average improvement of PPP-AR stability is 16.54%. Furthermore, using PPP-AR for time transfer not only enhances the stability of the time transfer link clock offset sequence, but also improves its noise level. The stability is increased by 14.41% on average, and the noise level is improved by 9.11% on average. PPP-RTK also improves the stability and noise level of time transfer even better than PPP-AR. However, as the reference network scale increases, the accuracy of the interpolated atmospheric delay correction decreases, impacting the performance of PPP-RTK.

MTFR: An universal multimodal fusion method through Modality Transfer and Fusion Refinement

Multimodal learning has made great achievements in recent years. However, most of the existing modal fusion methods capture cross-modal correlations in a single stage and are designed for specific tasks, with poor generalizability and portability. In this work, we propose the Modality Transfer and Fusion Refinement method (MTFR), which contains two feature-level fusion modules and can be used for several applications. Specifically, the Modality Transfer greatly shortens the distance between modalities through transferring non-text information into text-like representations. Fusion Refinement corrects and complements the previous fusion process to make the model focus more on task-specific knowledge by introducing uni-modal processing. To illustrate the validity of MTFR, we design three experiments for sarcasm detection, sentiment analysis, and offensiveness detection. The experimental results validate that our method achieves competitive performance in all three applications and outperforms both task-specific models and multimodal fusion baselines, which demonstrate the good generalizability and robustness of our method. Thanks to the promising performance, the method will facilitate the development of multimodal fusion and can promote multimodal applications in practice such as social sentiment analysis and multimodal conversational robots.

Highly sensitive and specific detection of human papillomavirus type 16 using CRISPR/Cas12a assay coupled with an enhanced single nanoparticle dark-field microscopy imaging technique

Human papillomavirus (HPV) is a prevalent sexually transmitted pathogen associated with cervical cancer. Detecting high-risk HPV (hr-HPV) infections is crucial for cervical cancer prevention, particularly in resource-limited settings. Here, we present a highly sensitive and specific sensor for HPV-16 detection based on CRISPR/Cas12a coupled with enhanced single nanoparticle dark-field microscopy (DFM) imaging techniques. Ag–Au satellites were assembled through the hybridization of AgNPs-based spherical nucleic acid (Ag-SNA) and AuNPs-based spherical nucleic acid (Au-SNA), and their disassembly upon target-mediated cleavage by the Cas12a protein was monitored using DFM for HPV-16 quantification. To enhance the cleavage efficiency and detection sensitivity, the composition of the ssDNA sequences on Ag-SNA and Au-SNA was optimized. Additionally, we explored using the SynSed technique (synergistic sedimentation of Brownian motion suppression and dehydration transfer) as an alternative particle transfer method in DFM imaging to traditional electrostatic deposition. This addresses the issue of inconsistent deposition efficiency of Ag–Au satellites and their disassembly due to their size and charge differences. The sensor achieved a remarkable limit of detection (LOD) of 10 fM, lowered by 9-fold compared to traditional electrostatic deposition methods. Clinical testing in DNA extractions from 10 human cervical swabs demonstrated significant response differences between the positive and negative samples. Our sensor offers a promising solution for sensitive and specific HPV-16 detection, with implications for cancer screening and management.

Controlling On-Surface Photoactivity: The Impact of π-Conjugation in Anhydride-Functionalized Molecules on a Semiconductor Surface.

On-surface synthesis has become a prominent method for growing low-dimensional carbon-based nanomaterials on metal surfaces. However, the necessity of decoupling organic nanostructures from metal substrates to exploit their properties requires either transfer methods or new strategies to perform reactions directly on inert surfaces. The use of on-surface light-induced reactions directly on semiconductor/insulating surfaces represents an alternative approach to address these challenges. Here, exploring the photochemical activity of different organic molecules on a SnSe semiconductor surface under ultra-high vacuum, we present a novel on-surface light-induced reaction. The selective photodissociation of the anhydride group is observed, releasing CO and CO2. Moreover, we rationalize the relationship between the photochemical activity and the π-conjugation of the molecular core. The different experimental behaviour of two model anhydrides was elucidated by theoretical calculations, showing how the molecular structure influences the distribution of the excited states. Our findings open new pathways for on-surface synthesis directly on technologically relevant substrates.