Non-negative Matrix Factorization Algorithm Research Articles

Comodule Estimation of Cognitive Sensor Networks Based on Partial Clustering for Partial Observed Data

The proposed study is on the partial clustering algorithms for cognitive sensor networks that deal with partially observed data. The proposed algorithms aim to estimate clusters in the presence of missing values and leverage data imputation techniques to fill in the gaps in the target and station device matrices. A modified loss function is introduced to shape the cluster centers, and robust Non-negative Matrix Factorization (NMF) algorithms are utilized to enhance the robustness of the clustering process. This research contributes to the field of cognitive sensor networks by providing insights into the challenges of partial clustering and presenting effective algorithms to address them. The proposed methods have the potential to enhance the performance of clustering tasks in various domains, including sensor networks, by accounting for missing data and producing accurate cluster reconstructions.

Adipose tissue composition determines its computed tomography radiodensity.

Adipose tissue radiodensity in computed tomography (CT) performed before surgeries can predict surgical difficulty. Despite its clinical importance, little is known about what influences radiodensity. This study combines desorption electrospray ionization mass spectrometry imaging (DESI-MSI) and electrospray ionization (ESI) with machine learning to unveil how chemical composition of adipose tissue determines its radiodensity. Patients in the study underwent abdominal surgeries. Before surgery, CT radiodensity of fat near operated sites was measured. Fifty-three fat samples were collected and analyzed by DESI-MSI, ESI, and histology, and then sorted by radiodensity, demographic parameters, and adipocyte size. A non-negative matrix factorization (NMF) algorithm was developed to differentiate between high and low radiodensities. No associations between radiodensity and patient age, gender, weight, height, or fat origin were found. Body mass index showed negative correlation with radiodensity. A substantial difference in chemical composition between adipose tissues of high and low radiodensities was observed. More radiodense tissues exhibited greater abundance of high molecular weight species, such as phospholipids of various types, ceramides, cholesterol esters and diglycerides, and about 70% smaller adipocyte size. Less radiodense tissue showed high abundance of short acyl-tail fatty acids. This study unveils the connection between abdominal adipose tissue radiodensity and its chemical composition. Because the radiodensity of the fat around the surgical site is associated with surgical difficulty, it is important to understand how adipose tissue composition affects this parameter. We conclude that fat tissue with a higher content of various phospholipids and waxy lipids is more CT radiodense. This study establishes the connection between the CT radiodensity of adipose tissue and its chemical composition. Clinicians may use this information for preoperative planning of surgical procedures, potentially modifying their surgical approach (for example, performing partial nephrectomy openly rather than laparoscopically). • Adipose tissue radiodensity values in computed tomography images taken prior to the surgery can potentially predict surgery difficulty. • Fifty-three human specimens were analyzed by advanced mass spectrometry, molecular imaging, and machine learning to establish the key features that determine Hounsfield units' values of adipose tissue. • The findings of this research will enable clinicians to better prepare for surgical procedures and select operative strategies.

Nonnegative Matrix Factorization: A Review

Recent developments in Non-negative Matrix Factorization (NMF) have focused on addressing several challenges and advancing its applicability. New algorithmic variations, such as robust NMF, deep NMF, and graph-regularized NMF, have emerged to improve NMF's performance in various domains. These developments aim to enhance the interpretability, scalability, and robustness of NMF-based solutions. NMF is now widely used in audio source separation, text mining, recommendation systems, and image processing. However, NMF still faces challenges, including sensitivity to initialization, the determination of the appropriate rank, and computational complexity. Overlapping sources in audio and data sparsity in some applications remain challenging issues. Additionally, ensuring the consistency and stability of NMF results in noisy environments is a subject of ongoing research. The quest for more efficient and scalable NMF algorithms continues, especially for handling large datasets. While NMF has made significant strides in recent years, addressing these challenges is crucial for unlocking its full potential in diverse data analysis and source separation tasks.

An Entropy Weighted Nonnegative Matrix Factorization Algorithm for Feature Representation.

Nonnegative matrix factorization (NMF) has been widely used to learn low-dimensional representations of data. However, NMF pays the same attention to all attributes of a data point, which inevitably leads to inaccurate representations. For example, in a human-face dataset, if an image contains a hat on a head, the hat should be removed or the importance of its corresponding attributes should be decreased during matrix factorization. This article proposes a new type of NMF called entropy weighted NMF (EWNMF), which uses an optimizable weight for each attribute of each data point to emphasize their importance. This process is achieved by adding an entropy regularizer to the cost function and then using the Lagrange multiplier method to solve the problem. Experimental results with several datasets demonstrate the feasibility and effectiveness of the proposed method. The code developed in this study is available at https://github.com/Poisson-EM/Entropy-weighted-NMF.

The Kinematic-Muscle Synergies during Infant Crawling: A Pilot Study.

Many feature extraction algorithms have been separately used for kinematic or muscle synergy analysis during human movement. However, very few studies focus on the co-extraction of kinematic and muscle synergies. Therefore, the aim of this study was to propose a novel and efficient approach for extracting the kinematic-muscle synergies during infant crawling. Surface electromyography signals and three-dimensional joint trajectories were collected from 20 typically developing infants during self-paced hands-and-knees crawling. Angular accelerations of shoulder, elbow, hip and knee flexion/extension computing from those joint trajectories were divided into two independent directional positive degrees-of-freedom. The kinematic-muscle synergies and corresponding activation coefficients were extracted using the non-negative matrix factorization algorithm based on two selection criteria of synergy number (i.e., criterion 1: the total constraint, criterion 2: a combination of the total constraint and a local constraint for each joint/muscle). Then, the data of each joint/muscle were reconstructed by those synergies and corresponding activation coefficients. Our results indicated that the minimum number of kinematic-muscle synergies based on criterion 1 is less than that based on criterion 2. The data reconstruction of joint flexion/extension based on criterion 2 is better than that based on criterion 1, whereas the data reconstruction of muscles is similar between criterion 1 and 2. These promising results show the feasibility of applying the proposed approach to clinical assessments of motor function for infants.Clinical Relevance- Extracting kinematic-muscle synergies during infant crawling has the potential for professional therapists or rehabilitation physicians to conduct the early assessment and rehabilitation treatment of infants with the central nervous system disorders.

A novel force-constrained non-negative matrix factorization algorithm reveals the effectiveness of muscle synergies in the task space.

According to the synergy hypothesis, the motor system recruits a small number of synergies in a task-dependent manner. Existing synergy extraction algorithms typically only consider the muscle pattern and it remains unclear to which extent muscle synergies encode task-relevant variations of muscle activity. We propose a novel force-constrained non-negative matrix algorithm (FCNMF) based on a gradient descent update rule that considers also the task space by adding a term penalizing force reconstruction error in the cost function. We validated the FCNMF algorithm using simulated muscle data and corrupted them by noise. We compared task performances with reconstructed trajectories using synergies (RS) extracted from the FCNMF algorithm and from the standard multiplicative non-negative matrix factorization NMF algorithm. We found that FCNMF outperforms NMF for different types of noise. Finally, we demonstrated the effectiveness of FCNMF on EMG data collected during an isometric reaching task. The new algorithm accurately reconstructs the trajectories in all participants, even in those for which the NMF algorithm fails. These findings show the effectiveness of muscle synergies extracted considering the task space, possibly thanks to the robustness of FCNMF against non-isotropic noise present in muscle data, suggesting that they provide an effective strategy for motor coordination.

Genome-scale CRISPR-Cas9 screening stratifies pancreatic cancer with distinct outcomes and immunotherapeutic efficacy

Pancreatic cancer (PC) was featured by dramatic heterogeneity and dismal outcomes. An ideal classification strategy capable of achieving risk stratification and individualized treatment is urgently needed to significantly improve prognosis. In this study, using the 105 prognostic cancer essential genes identified by genome-scale CRISPR-Cas9 screening and univariate Cox analysis, we established and verified three heterogeneous subtypes via non-negative matrix factorization (NMF) and nearest template prediction (NTP) algorithms in the TCGA-PAAD cohort (176 samples) and four multi-center cohorts (233 samples), respectively. Among them, C1 with the worst prognosis was enriched in immune-related pathways, possessed superior infiltration abundance of immune cells and immune checkpoint molecules expression, and might be most sensitive to immunotherapy. C3, owing a moderate prognosis, might be featured by proliferative biological function, and despite its highest immunogenicity, the defects in antigen processing and presentation ability coupled with barren immune environment render it ineffective for immunotherapy, while it had potential sensitivity to paclitaxel and methotrexate. Besides, C2 harbored the best prognosis and was characterized by metabolism-related functions. These results could deepen our understanding of PC molecular heterogeneity and provide a trustworthy reference for prognostic stratification management and precision medicine in clinical practice.

Metabolism-related gene vaccines and immune infiltration in ovarian cancer: A novel risk score model of machine learning.

The present study aims to develop a metabolic gene signature to evaluate the survival rate of ovarian cancer (OC) patients and analyze the potential mechanisms of metabolic genes in OC because the difficulty in early detection of OC often leads to poor treatment outcomes. A non-negative matrix factorization algorithm was applied to determine molecular subtypes according to metabolism genes. To build a risk prognosis model, least absolute shrinkage and selection operator multivariate Cox analysis was carried out with weighted correlation network analysis (WCGNA). Glycolytic flux and mitochondrial function were evaluated by conducting seahorse analysis. On the basis of metabolism-related genes, the two subtypes of OC samples present in The Cancer Genome Atlas database were distinguished. An analysis of WGCNA identified 1056 genes. Lastly, a 10-gene signature (CMAS, ADH1B, PLA2G2D, BHMT, CACNA1C, AADAC, ALOX12, CYP2R1, SCN1B and ME1) was constructed that demonstrated promising performance in predicting outcome in patients with OC. The RiskScore of the gene signature was linked to microenvironment cell infiltration and immune checkpoint. Higher RiskScores were associated with poorer results for OC patients. Seahorse analysis shows the influence of CMAS in cell energy metabolism. In the present study, a novel marker for evaluating the survival of OC patients was developed through the creation of a gene signature incorporating metabolism-related genes. Our knowledge of immunotherapy and microenvironment cell infiltration may be enriched by evaluating metabolism-related gene modification patterns.

Identification of N1 methyladenosine-related biomarker predicting overall survival outcomes and experimental verification in ovarian cancer.

This study aimed to construct a N1-methyladenosine (m1A)-related biomarker model for predicting the prognosis of ovarian cancer (OVCA). OVCA samples were clustered into two subtypes using the Non-Negative Matrix Factorization (NMF) algorithm, including TCGA (n = 374) as the training set and GSE26712 (n = 185) as the external validation set. Hub genes, which were screened to construct a risk model, and nomogram to predict the overall survival of OVCA were explored and validated through various bioinformatic analysis and quantitative real-time PCR. Following bootstrap correction, the C-index of nomogram was 0.62515, showing reliable performance. The functions of DEGs in the high- and low-risk groups were mainly enriched in immune response, immune regulation, and immune-related diseases. The immune cells relevant to the expression of hub genes were explored, for example, Natural Killer (NK) cells, T cells, activated dendritic cells (aDC). AADAC, CD38, CACNA1C, and ATP1A3 might be used as m1A-related biomarkers for OVCA, and the nomogram labeled with m1A for the first time had excellent performance for predicting overall survival in OVCA.

Deep Multiplicative Update Algorithm for Nonnegative Matrix Factorization and Its Application to Audio Signals

Deep Multiplicative Update Algorithm for Nonnegative Matrix Factorization and Its Application to Audio Signals

Robustness of NMF Algorithms Under Different Noises

In machine learning, datasets are often disturbed by different noises. The Nonnegative Matrix Factorization (NMF) algorithm provides a robust method to deal with noise, which will significantly improve the efficiency of machine learning. In this investigation, the standard NMF algorithm and L2,1-Norm Based NMF algorithm are studied by designing experiments on different noise types, noise levels, and datasets. Furthermore, Relative Reconstruction Errors (RRE), accuracy, and Normalized Mutual Information (NMI) are used to evaluate the robustness of the two algorithms. In this experiment, there is no significant difference in performance between the two algorithms, while L2,1-Norm Based NMF algorithm shows relatively small advantages.

Metabolic Pathway Pairwise-Based Signature as a Potential Non-Invasive Diagnostic Marker in Alzheimer's Disease Patients.

Alzheimer's disease (AD) is an incurable neurodegenerative disorder. Early screening, particularly in blood plasma, has been demonstrated as a promising approach to the diagnosis and prevention of AD. In addition, metabolic dysfunction has been demonstrated to be closely related to AD, which might be reflected in the whole blood transcriptome. Hence, we hypothesized that the establishment of a diagnostic model based on the metabolic signatures of blood is a workable strategy. To that end, we initially constructed metabolic pathway pairwise (MPP) signatures to characterize the interplay among metabolic pathways. Then, a series of bioinformatic methodologies, e.g., differential expression analysis, functional enrichment analysis, network analysis, etc., were used to investigate the molecular mechanism behind AD. Moreover, an unsupervised clustering analysis based on the MPP signature profile via the Non-Negative Matrix Factorization (NMF) algorithm was utilized to stratify AD patients. Finally, aimed at distinguishing AD patients from non-AD groups, a metabolic pathway-pairwise scoring system (MPPSS) was established using multi-machine learning methods. As a result, many metabolic pathways correlated to AD were disclosed, including oxidative phosphorylation, fatty acid biosynthesis, etc. NMF clustering analysis divided AD patients into two subgroups (S1 and S2), which exhibit distinct activities of metabolism and immunity. Typically, oxidative phosphorylation in S2 exhibits a lower activity than that in S1 and non-AD group, suggesting the patients in S2 might possess a more compromised brain metabolism. Additionally, immune infiltration analysis showed that the patients in S2 might have phenomena of immune suppression compared with S1 and the non-AD group. These findings indicated that S2 probably has a more severe progression of AD. Finally, MPPSS could achieve an AUC of 0.73 (95%CI: 0.70, 0.77) in the training dataset, 0.71 (95%CI: 0.65, 0.77) in the testing dataset, and an AUC of 0.99 (95%CI: 0.96, 1.00) in one external validation dataset. Overall, our study successfully established a novel metabolism-based scoring system for AD diagnosis using the blood transcriptome and provided new insight into the molecular mechanism of metabolic dysfunction implicated in AD.

Identification and validation of Birc5 as a novel activated cell cycle program biomarker associated with infiltration of immunosuppressive myeloid-derived suppressor cells in hepatocellular carcinoma.

Preclinical studies and clinical trials have demonstrated that tumor-intrinsic activation of the cell cycle program impedes anticancer immunotherapy. Identification of cell cycle-related biomarkers may provide novel therapeutic targets to augment the efficacy of immunotherapy in hepatocellular carcinoma (HCC). Based on the genes related to cell cycle program, two clusters (Cluster 1 and Cluster 2) were detected in HCC patients via non-negative matrix factorization algorithm. Multivariable-adjusted Cox regression analysis indicated that the cell cycle gene-based classification was a significant prognostic factor for predicting the clinical outcome of HCC patients. Cluster 1 showed shorter overall survival time and progression-free interval time was associated with activated cell cycle program, higher infiltration of myeloid-derived suppressor cells (MDSCs) and less sensitivity to immunotherapy. A three-gene prognostic model, including BIRC5, C8G, and SPP1, was constructed to characterize the cell cycle-based classification of HCC, which had strong robustness and a stable predictive performance. Notably, Birc5 was positively correlated with CD11b expression (a MDSC marker) in HCC tissue. Concordant high expression of Birc5 and intratumor infiltration level of MDSCs were correlated with worse prognosis of HCC patients. In vitro, hepatocellular Birc5 overexpression promoted immunosuppressive CD11b+ CD33+ HLA-DR- MDSC expansion from human peripheral blood mononuclear cells. Genetically modified animal model of liver cancer revealed that Birc5 depletion upregulated the genes related to lymphocyte-mediated immunity, natural killer cell-mediated immunity, interferon-gamma production, T-cell activation, and T-cell-mediated cytotoxicity. These results suggest an immunosuppressive function of Birc5 in HCC. Birc5 was a potential biomarker and inducer of intratumor infiltration of MDSCs, which led to T cell exclusion or dysfunction in tumor immune microenvironment, consequently resulting in reduced response to ICIs in HCC.

Enhanced Extraction of Blood and Tissue Time-Activity Curves in Cardiac Mouse FDG PET Imaging by Means of Constrained Nonnegative Matrix Factorization.

We propose an enhanced method to accurately retrieve time-activity curves (TACs) of blood and tissue from dynamic 2-deoxy-2-[18F]fluoro-D-glucose ([18F]FDG) positron emission tomography (PET) cardiac images of mice. The method is noninvasive and consists of using a constrained nonnegative matrix factorization algorithm (CNMF) applied to the matrix (A) containing the intensity values of the voxels of the left ventricle (LV) PET image. CNMF factorizes A into nonnegative matrices H and W, respectively, representing the physiological factors (blood and tissue) and their associated weights, by minimizing an extended cost function. We verified our method on 32 C57BL/6 mice, 14 of them with acute myocardial infarction (AMI). With CNMF, we could break down the mouse LV into myocardial and blood pool images. Their corresponding TACs were used in kinetic modeling to readily determine the [18F]FDG influx constant (Ki) required to compute the myocardial metabolic rate of glucose. The calculated Ki values using CNMF for the heart of control mice were in good agreement with those published in the literature. Significant differences in Ki values for the heart of control and AMI mice were found using CNMF. The values of the elements of W agreed well with the LV structural changes induced by ligation of the left coronary artery. CNMF was compared with the recently published method based on robust unmixing of dynamic sequences using regions of interest (RUDUR). A clear improvement of signal separation was observed with CNMF compared to the RUDUR method.

Graph Regularized Sparse Non-Negative Matrix Factorization for Clustering

The graph regularized nonnegative matrix factorization (GNMF) algorithms have received a lot of attention in the field of machine learning and data mining, as well as the square loss method is commonly used to measure the quality of reconstructed data. However, noise is introduced when data reconstruction is performed; and the square loss method is sensitive to noise, which leads to degradation in the performance of data analysis tasks. To solve this problem, a novel graph regularized sparse NMF (GSNMF) is proposed in this article. To obtain a cleaner data matrix to approximate the high-dimensional matrix, the l₁-norm to the low-dimensional matrix is added to achieve the adjustment of data eigenvalues in the matrix and sparsity constraint. In addition, the corresponding inference and alternating iterative update algorithm to solve the optimization problem are given. Then, an extension of GSNMF, namely, graph regularized sparse nonnegative matrix trifactorization (GSNMTF), is proposed, and the detailed inference procedure is also shown. Finally, the experimental results on eight different datasets demonstrate that the proposed model has a good performance.

Muscle synergy analysis of eight inter-limb coordination modes during human hands-knees crawling movement.

In order to reveal in-depth the neuromuscular control mechanism of human crawling, this study carries out muscle synergy extraction and analysis on human hands-knees crawling under eight specific inter-limb coordination modes, which are defined according to the swing sequence of limbs and includes two-limb swing crawling modes and six single-limb swing crawling modes. Ten healthy adults participate in crawling data collection, and surface electromyography (sEMG) signals are recorded from 30 muscles of limbs and trunk. Non-negative matrix factorization (NNMF) algorithm is adopted for muscle synergy extraction, and a three-step muscle synergy analysis scheme is implemented by using the hierarchical clustering method. Based on results of muscle synergy extraction, 4 to 7 synergies are extracted from each participant in each inter-limb coordination mode, which supports the muscle synergy hypothesis to some extent, namely, central nervous system (CNS) controls the inter-limb coordination modes during crawling movement by recruiting a certain amount of muscle synergies, rather than a single muscle. In addition, when different participants crawl in the same inter-limb coordination mode, they share more temporal features in recruiting muscle synergies. Further, by extracting and analyzing intra-mode shared synergies among participants and inter-mode shared synergies among the eight inter-limb coordination modes, the CNS is found to realize single-limb swing crawling modes by recruiting the four inter-mode shared synergy structures related to the swing function of each limb in different orders, and realize the two-limb swing crawling modes by recruiting synchronously two intra-mode shared synergy structures. The research results of the muscle synergy analysis on the eight specific inter-limb coordination modes, on the one hand, provide a basis for muscle synergy hypothesis from the perspective of crawling motion, on the other hand, also provide a possible explanation for the choice of the inter-limb coordination mode in human crawling.

Nonnegative matrix factorization analysis and multiple machine learning methods identified IL17C and ACOXL as novel diagnostic biomarkers for atherosclerosis

BackgroundAtherosclerosis is the common pathological basis for many cardiovascular and cerebrovascular diseases. The purpose of this study is to identify the diagnostic biomarkers related to atherosclerosis through machine learning algorithm.MethodsClinicopathological parameters and transcriptomics data were obtained from 4 datasets (GSE21545, GSE20129, GSE43292, GSE100927). A nonnegative matrix factorization algorithm was used to classify arteriosclerosis patients in GSE21545 dataset. Then, we identified prognosis-related differentially expressed genes (DEGs) between the subtypes. Multiple machine learning methods to detect pivotal markers. Discrimination, calibration and clinical usefulness of the predicting model were assessed using area under curve, calibration plot and decision curve analysis respectively. The expression level of the feature genes was validated in GSE20129, GSE43292, GSE100927.Results2 molecular subtypes of atherosclerosis was identified, and 223 prognosis-related DEGs between the 2 subtypes were identified. These genes are not only related to epithelial cell proliferation, mitochondrial dysfunction, but also to immune related pathways. Least absolute shrinkage and selection operator, random forest, support vector machine- recursive feature elimination show that IL17C and ACOXL were identified as diagnostic markers of atherosclerosis. The prediction model displayed good discrimination and good calibration. Decision curve analysis showed that this model was clinically useful. Moreover, IL17C and ACOXL were verified in other 3 GEO datasets, and also have good predictive performance.ConclusionIL17C and ACOXL were diagnostic genes of atherosclerosis and associated with higher incidence of ischemic events.

Identification of lung adenocarcinoma subtypes and predictive signature for prognosis, immune features, and immunotherapy based on immune checkpoint genes.

Background: Lung adenocarcinoma (LUAD) is the most common variant of non-small cell lung cancer (NSCLC) across the world. Recently, the rapid development of immunotherapy has brought a new dawn for LUAD patients. Closely related to the tumor immune microenvironment and immune cell functions, more and more new immune checkpoints have been discovered, and various cancer treatment studies targeting these novel immune checkpoints are currently in full swing. However, studies on the phenotype and clinical significance of novel immune checkpoints in LUAD are still limited, and only a minority of patients with LUAD can benefit from immunotherapy. Methods: The LUAD datasets were downloaded from The Cancer Genome Atlas (TCGA) and the Gene Expression Omnibus (GEO) databases, and the immune checkpoints score of each sample were calculated based on the expression of the 82 immune checkpoints-related genes (ICGs). The weighted gene co-expression network analysis (WGCNA) was used to obtain the gene modules closely related to the score and two different LUAD clusters were identified based on these module genes by the Non-negative Matrix Factorization (NMF) Algorithm. The differentially expressed genes between the two clusters were further used to construct a predictive signature for prognosis, immune features, and the response to immunotherapy for LUAD patients through a series of regression analyses. Results: A new immune checkpoints-related signature was finally established according to the expression of 7 genes (FCER2, CD200R1, RHOV, TNNT2, WT1, AHSG, and KRTAP5-8). This signature can stratify patients into high-risk and low-risk groups with different survival outcomes and sensitivity to immunotherapy, and the signature has been well validated in different clinical subgroups and validation cohorts. Conclusion: We constructed a novel immune checkpoints-related LUAD risk assessment system, which has a good predictive ability and significance for guiding immunotherapy. We believe that these findings will not only aid in the clinical management of LUAD patients but also provide some insights into screening appropriate patients for immunotherapy.

A Generalized Deep Learning Clustering Algorithm Based on Non-Negative Matrix Factorization

Clustering is a popular research topic in the field of data mining, in which the clustering method based on non-negative matrix factorization (NMF) has been widely employed. However, in the update process of NMF, there is no learning rate to guide the update as well as the update depends on the data itself, which leads to slow convergence and low clustering accuracy. To solve these problems, a generalized deep learning clustering (GDLC) algorithm based on NMF is proposed in this article. Firstly, a nonlinear constrained NMF (NNMF) algorithm is constructed to achieve sequential updates of the elements in the matrix guided by the learning rate. Then, the gradient values corresponding to the element update are transformed into generalized weights and generalized biases, by inputting the elements as well as their corresponding generalized weights and generalized biases into the nonlinear activation function to construct the GDLC algorithm. In addition, for improving the understanding of the GDLC algorithm, its detailed inference procedure and algorithm design are provided. Finally, the experimental results on eight datasets show that the GDLC algorithm has efficient performance.

Identification of anoikis-related subtypes and development of risk stratification system in skin cutaneous melanoma

BackgroundAnoikis, a form of apoptosis induced by cell detachment, plays a key role in cancer metastasis. However, the potential roles of anoikis-related genes (ARGs) in assessing the prognosis of skin cutaneous melanoma (SKCM) and the tumor microenvironment (TME) remain unclear. MethodsThe data from TCGA corresponding to transcriptomic expression patterns for patients with SKCM were downloaded and utilized to screen distinct molecular subtypes by a non-negative matrix factorization algorithm. The prognostic signature was constructed by least absolute shrinkage and selection operator (LASSO) Cox regression and was validated in SKCM patients from the GEO cohort. Moreover, the relationship of the ARG_score with prognosis, tumor-infiltrating immune cells, gene mutation, microsatellite instability (MSI), and immunotherapy efficacy. ResultsWe screened 100 anoikis-related differentially expressed genes between SKCM tissues and normal skin tissues, which could divide all patients into three different subtypes with significantly distinct prognosis and immune cell infiltration. Then, an anoikis-related signature was developed based on subtype-related DEGs, which could classify all SKCM patients into low and high ARG_score groups with differing overall survival (OS) rates. ARG_score was confirmed to be a strong independent prognostic indicator for SKCM patients. By combining ARG_score with clinicopathological features, a nomogram was constructed, which could accurately predict the individual OS of patients with SKCM. Moreover, low ARG_score patients presented with higher levels of immune cell infiltration, TME score, higher tumor mutation burden, and better immunotherapy responses. ConclusionsOur comprehensive analysis of ARGs in SKCM provides important insights into the immunological microenvironment within the tumor of SKCM patients and helps to forecast prognosis and the response to immunotherapy in SKCM patients, thereby making it easier to tailor more effective treatment strategies to individual patients.