RefLaTEA: a robust visualization and analysis framework leveraging background data for enhanced insight.

Hierarchical density-based spatial clustering of applications with noise (HDBSCAN) and uniform manifold approximation and projection (UMAP), two new state-of-the-art algorithms for clustering analysis, and dimensionality reduction, respectively, are proposed for the segmentation of core-loss electron energy loss spectroscopy (EELS) spectrum images. The performances of UMAP and HDBSCAN are systematically compared to the other clustering analysis approaches used in EELS in the literature using a known synthetic dataset. Better results are found for these new approaches. Furthermore, UMAP and HDBSCAN are showcased in a real experimental dataset from a core–shell nanoparticle of iron and manganese oxides, as well as the triple combination nonnegative matrix factorization–UMAP–HDBSCAN. The results obtained indicate how the complementary use of different combinations may be beneficial in a real-case scenario to attain a complete picture, as different algorithms highlight different aspects of the dataset studied.

ABSTRACTMicrobiome data are sparse and high dimensional, so effective visualization of these data requires dimensionality reduction. To date, the most commonly used method for dimensionality reduction in the microbiome is calculation of between-sample microbial differences (beta diversity), followed by principal-coordinate analysis (PCoA). Uniform Manifold Approximation and Projection (UMAP) is an alternative method that can reduce the dimensionality of beta diversity distance matrices. Here, we demonstrate the benefits and limitations of using UMAP for dimensionality reduction on microbiome data. Using real data, we demonstrate that UMAP can improve the representation of clusters, especially when the clusters are composed of multiple subgroups. Additionally, we show that UMAP provides improved correlation of biological variation along a gradient with a reduced number of coordinates of the resulting embedding. Finally, we provide parameter recommendations that emphasize the preservation of global geometry. We therefore conclude that UMAP should be routinely used as a complementary visualization method for microbiome beta diversity studies.IMPORTANCE UMAP provides an additional method to visualize microbiome data. The method is extensible to any beta diversity metric used with PCoA, and our results demonstrate that UMAP can indeed improve visualization quality and correspondence with biological and technical variables of interest. The software to perform this analysis is available under an open-source license and can be obtained at https://github.com/knightlab-analyses/umap-microbiome-benchmarking; additionally, we have provided a QIIME 2 plugin for UMAP at https://github.com/biocore/q2-umap.

Introduction: Accurate diagnosis is essential for cancer treatment. Sometimes, it is difficult to confirm histological diagnosis due to inadequate tissue sample or preparation. Using molecular markers for classification would be ideal in such cases. For renal cell carcinoma(RCC), we developed a molecular classification based on transcriptomics to address these issues and aid current diagnostic workflow. Method: We obtained transcriptomic and phenotype data of renal cell carcinoma from The cancer genome atlas (TCGA) data repository. We developed an unsupervised algorithm, Density-based UMAP, based on two clustering methods, UMAP (Uniform manifold approximation and projection) and DBScan (Density-based spatial clustering of applications with noise). We iterated this algorithm 1000 times with random 1000 genes each time and classified each sample in each iteration. We used plurality voting of at least 70% of the iterations for consensus groups. The algorithm was able to identify all the major histological subtypes. We ran a differential expression analysis for each group. Finally, we developed a classification gene signature and a supervised algorithm to classify subtypes with fewer genes and implement this signature in an RCC metanalysis. Results: Density-based UMAP Algorithm classified the samples with 91.4% concordance with WHO 2016 classification. Among the discrepant cases, 4.4 % cases (46) were not diagnosed the same as their histological classification. 4.2 % (44) cases were shown a mixed expression profile. These cases are of further interest as they might respond better with a different treatment. Based on molecular characteristics they share with major subtypes, they are named ClCh, Clear 2, and Clear 3. ClCh and Clear 2 have shown to be indolent and, Clear 3 has shown to be very aggressive. We manually curated 328 genes for supervised learning from differential expression analysis between the groups and able to identify the groups in a metanalysis. Conclusion: Transcriptomic profiling is fast, robust and, simple. It can help in diagnosing histologically discrepant cases. This classifier would be a good edition in renal cancer diagnostic workflow and can complement current diagnostic guidelines for renal cancer. Citation Format: Khaled Bin Satter, Paul MH Tran, Sharad B. Purohit, Jin-Xiong She. Transcriptomic classification of renal cancer: a machine learning approach [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2021; 2021 Apr 10-15 and May 17-21. Philadelphia (PA): AACR; Cancer Res 2021;81(13_Suppl):Abstract nr 2192.

Simple SummaryRecognizing the different species can help us better understand the nature. One way to differentiate bird species is the bird song. There are mathematical techniques that extract information from the bird songs, potentially allowing automatic differentiation of species. However, there is still a lack of techniques that use the extracted information and accurately differentiate individuals of different species. For the first time, we have used a technique called Uniform Manifold Approximation and Projection (UMAP) to identify the two taxonomic groups of a bird named Rough-legged Tyrannulet Phyllomyias burmeisteri, which is a species that can be found all the way from Costa Rica to Argentina. Although there is evidence of the existence of two taxonomic groups, previous studies have shown them to be difficult to distinguish. We collected Rough-legged Tyrannulet bird songs of 101 birds from 11 countries. UMAP allowed us to make a transformation of the multiple measures obtained from the bird song of each bird into just two values. Plotting the UMAP values of each bird in a two-dimensional graph, it turns out that UMAP was able to clearly identify the two taxonomic groups, which has been named as Rough-legged Tyrannulet Phyllomyias burmeisteri and White-fronted Tyrannulet Phyllomyias zeledoni. UMAP can potentially help the identification of other species difficult to classify.Vocalizations from birds are a fruitful source of information for the classification of species. However, currently used analyses are ineffective to determine the taxonomic status of some groups. To provide a clearer grouping of taxa for such bird species from the analysis of vocalizations, more sensitive techniques are required. In this study, we have evaluated the sensitivity of the Uniform Manifold Approximation and Projection (UMAP) technique for grouping the vocalizations of individuals of the Rough-legged Tyrannulet Phyllomyias burmeisteri complex. Although the existence of two taxonomic groups has been suggested by some studies, the species has presented taxonomic difficulties in classification in previous studies. UMAP exhibited a clearer separation of groups than previously used dimensionality-reduction techniques (i.e., principal component analysis), as it was able to effectively identify the two taxa groups. The results achieved with UMAP in this study suggest that the technique can be useful in the analysis of species with complex in taxonomy through vocalizations data as a complementary tool including behavioral traits such as acoustic communication.

Dear editors, Despite efforts to implement vaccination and screening programs, cervical cancer (CC) remains a public health problem [1]. Over 90% of cases are associated with persistent human papillomavirus (HPV) infections. HPV integrates its DNA into the basal cells of the transformation zone, the region where the epithelium transitions from the columnar epithelial cells of the endocervix to the stratified squamous epithelium cells of the ectocervix, leading to the production of proteins (E6 and E7) that eventually cause dysplasia [2, 3]. Squamous cell carcinoma (SCC) and cervical adenocarcinoma (AD) are the two most common types of HPV-related CC [4]. SCC and AD have a long-term stage of preinvasive (PRE) lesion, which are named high-grade squamous intraepithelial lesion (HSIL) or adenocarcinoma in situ (AIS), respectively [5]. However, no studies were conducted to determine why some individuals with HPV infections develop HSIL while others develop AIS. To reveal the different characteristics of normal cervical progression to HSIL or AIS, we performed single-cell RNA-sequencing (scRNA-seq) on 8 samples, including 2 HPV-positive (HPV-P) normal cervical, 2 HPV-negative (HPV-N) normal cervical, 2 HSIL (PRE-HSIL) and 2 AIS (PRE-AIS) samples (Supplementary Table S1). The study protocols can be found in the Supplementary Materials. A total of 80,238 cells were obtained, with an average gene number of 2,260. Then, we acquired 33 clusters (Supplementary Figure S1A), which could be divided into 10 major cell types through the expression of typical marker genes (Supplementary Figure S1B-C). Based on the comparison of cell numbers, we found that, compared to HPV-N, HPV-P led to an increase in immune cells, such as B cells, natural killer (NK) cells/T cells, and neutrophils (Supplementary Figure S1D). When the lesion progressed to a preinvasive lesion, the prominent characteristic was the increase of epithelial cells in both AIS and HSIL. To determine the different cellular origins of AIS and HSIL, we reclustered the extracted 28,364 epithelial cells and acquired 13 clusters. The split Uniform Manifold Approximation and Projection (UMAP) map revealed that most epithelial cells in cluster 3 were from AIS, whereas those in clusters 4 and 5 were from HSIL (Figure 1A-B). Surprisingly, epithelial cells in clusters 3, 4 and 5 were found in lower numbers in the HPV-N or HPV-P samples, indicating that these cells represented the specific subpopulations of AIS and HSIL, respectively. Gene Ontology (GO) analysis revealed that cells in cluster 3 expressed high levels of cell cycle-related signaling pathways, such as the mitotic cell cycle and chromosome condensation (Supplementary Figure S2). Cells in cluster 4 were related to epithelial cell differentiation and actin cytoskeleton organization, whereas cells in cluster 5 exhibited humoral immune response, negative regulation of endopeptidase activity, inflammatory response and regulation of apoptotic signaling pathway (Supplementary Figure S2). Single-cell transcriptomics identifies the origin and immune microenvironment of cervical precancerous lesions. (A) UMAP showing that the number of epithelial cells in cluster 3 was higher in AIS and epithelial cells in cluster 4 and 5 were higher in HSIL compared to HPV-P and HPV-N group. Each dot represents a single cell, color-coded by cell clusters. X- (UMAP 1) and Y-axis (UMAP 2) represented the location of epithelial cells in the four groups after dimension reduction and unsupervised clustering (B) Bar graph showing the proportion of the 13 epithelial cell clusters quantified in four groups. (C) 2D graph of the pseudotime-ordered epithelial cells from HPV-N, HPV-P, PRE-AIS and PRE-HSIL samples, respectively. In the top panel, X-axis indicates the pseudotime, and Y-axis indicates the cell density distribution along the trajectory. In the bottom panel, X- (Component 1) and Y-axis (Component 2) label represented the location of epithelial cells on the pseudotime trajectories in the four groups. In the right panel, the branch trajectory plot with dark blue to light blue colors indicated the pseudotime of single cells. (D) Heatmap showing the dynamic changes in gene expression along the pseudotime. (E) Bubble plot showing the expression of markers genes in epithelial cells in the five cell types. The fraction of cells expressing genes was indicated by the size of the circle, and their scaled expression levels were indicated by the color of the circle. (F) UMAP of five types of epithelial cells. X- (UMAP 1) and Y-axis (UMAP 2) represented the location of five epithelial cell types. (G) Heatmap showing the enriched hallmark gene signatures of epithelial cells in five epithelial cell types. (H) The proportion of the five epithelial cell types in the four groups. The green arrow indicated the proportion of undefined Epi (I) UMAP of 13 NK/T cell clusters. X- (UMAP 1) and Y-axis (UMAP 2) represented the location of NK/T cells. (J) The proportion of 13 NK/T cell types in the four groups. (K) Boxplots comparing cytotoxicity, exhausted, naïve, proliferative, regulatory scores for NK/T cells between HPV-N, HPV-P, PRE-AIS and PRE-HSIL. (***P < 0.001; n = 10, 512). (L) Bubble plot showing the expression of the representative genes of NK/T cells in the four groups. The fraction of cells expressing genes was indicated by the size of the circle, and their scaled expression levels were indicated by the color of the circle. (M) Heatmap depicting the enriched hallmark gene signatures of NK/T cells in the four groups. (N) Schematic illustration of the development of human cervical AIS and HSIL. Abbreviations: UMAP, uniform manifold approximation and projection; AIS, adenocarcinoma in situ; HSIL, high-grade squamous intraepithelial lesion; HPV-P, human papillomavirus-positive normal cervix; HPV-N, human papillomavirus-negative normal cervix; 2D, 2-ddemension; PRE, preinvasive; NK, natural killer; To reveal the functional changes during lesion progression, we projected all epithelial cells onto pseudotime trajectories (Figure 1C). This unsupervised approach identified continuous cell states and formed two distinct trajectories, starting from state 1 and gradually progressing toward states 2 and 3, revealing a common origin with divergent fates (Figure 1C and Supplementary Figure S3). State 1 mainly comprised epithelial cells from HPV-N and HPV-P normal cervix, which were then separated into two branches. One branch described the developmental pathway of cells derived mainly from HSIL (cell fate 1, state 2), whereas the other branch was occupied by cells from AIS (cell fate 2, state 3). The branched heatmap showed that significant genes could be divided into four different clusters according to the gene expression dynamics (Figure 1D). GO analyses revealed that the highly expressed genes (gene set 2) in cell fate 1 were related to keratinocyte differentiation, epidermis development, and epidermal cell differentiation (Pathway 2). Meanwhile, immune regulation-related pathways, such as leukocyte-mediated immunity, complement activation and phagocytosis, were up-regulated from the normal cervix to the HSIL (Pathway 3). The highly expressed genes (gene set 1) in cell fate 2 had high enrichment of response to ketone, cellular response to metal ion and RNA splicing (Pathway 1). Together, these results demonstrated the different functional characteristics of epithelial cells between HSIL and AIS. Based on top differentially expressed genes (DEGs) and functional characteristics, we classified all epithelial cells (Epi) into five types: proliferative Epi (C3), differentiated Epi (C4, 6, and 10), inflammatory Epi (C2 and 7), metabolic Epi (C0, 1, 5, 8, 9, and 11) and undefined Epi (C12) (Figure 1E-F). Gene Set Variation Analysis (GSVA) confirmed their specific functional characteristics (Figure 1G). We compared the percentages of the five epithelial types in different groups and found that AIS had the highest percentage of proliferative Epi cells, while HSIL had the highest percentage of differentiated Epi cells (Figure 1H). Many studies have reported that epithelial-mesenchymal transition (EMT) is closely related to SCC progression and metastasis [6]. Herein, we found that differentiated Epi cells had a higher EMT score than other cell types (Supplementary Figure S4A-B). Similarly, cells in cluster 4 had the highest EMT score compared with other epithelial cell clusters (Supplementary Figure S4C), indicating that EMT played a key role in regulating the progression of lesions from HPV infection to HSIL. In our study, we identified 13 clusters (Supplementary Figure S5A-B), including 6 CD8+ T cell clusters (C0, 2, 3, 4, 10, and 12), 3 CD4+ T cell clusters (C1, 6, and 8), and 4 NK cell clusters (C5, 7, 9, and 11) (Figure 1I and Supplementary Figure S5C-E). Different groups had different percentages of 13 clusters (Figure 1J). Our results showed that, compared to the HPV-N cervix, the HPV-P cervix had a higher infiltration of cytotoxicity CD8+ T cells and effector memory CD8+ T cells but a lower infiltration of naïve CD4+ T cells. Similarly, HPV-P had higher cytotoxic and exhausted scores but lower naïve scores than HPV-N (Figure 1K), indicating that HPV infection could trigger an immune response. Compared with HSIL, AIS was associated with higher percentages of cytotoxic CD8+ T cells and exhausted CD8+ T cells, but a lower percentage of NK cells, indicating that both types were dominated by different effector cells. HSIL had the highest cytotoxic and exhausted scores compared to other groups (Figure 1K), indicating that HSIL had the best response rate to immunotherapy. More importantly, AIS and HSIL strongly expressed various immune-related genes and proliferative signaling pathways (Figure 1L-M), indicating that abnormal proliferation of epithelial cells contributed to the complexity of the microenvironment. Then, we performed a pseudotime analysis to understand the underlying evolution of the cellular status of CD8+ T, CD4+ T and NK cells (Supplementary Figure S6A-C). Interestingly, by combining the clustering and pseudotime analyses findings, we observed a gradual transition of CD8+ T and NK cells towards the subpopulations with immune enhancement from HPV infection to HSIL or AIS. Such an activation status for immune cells was indicated by the upregulation of GNLY, GZMB, FCER1G, and NKG7 and the downregulation of IL7R, CD69, GPR183 and NR4A1 in HSIL and AIS (Supplementary Figure S6D-E). More importantly, HSIL and AIS had similar immune activation states, which were characterized by the high infiltration of proliferative NK cells, regulatory CD4+ T cells, cytotoxic CD8+ T cells and exhausted CD8+ T cells. However, compared with AIS, HSIL was associated with higher infiltration of GNLY+ NK cells, naïve NK cells, and effector memory CD8+ T cells (Supplementary Figure S7). Our findings suggested that AIS might originate from proliferative epithelial cells, whereas HSIL might originate from differentiated epithelial cells. Further, HPV infection could affect the immune microenvironment, and HSIL and AIS exhibited different characteristics of immune activation (Figure 1N). Collectively, the results of this study provide new insights into the progression of normal cervix to the AIS and HSIL stages after HPV infection. Chunbo Li and Keqin Hua designed this study; Chunbo Li collected the samples, performed the experiments, analyzed the data, and drafted the manuscript. All authors participated in the writing and have read and approved the final manuscript. The authors would like to thank the patients, data managers, and all study investigators for their contributions to the study. The authors declare that they have no competing interests. This study was supported by the National Natural Science Foundation of China (Grant No. 82173188) to Keqin Hua, the Clinical Research Plan of Shanghai Hospital Development Central (SHDC) (SHDC2020CR1045B, SHDC2020CR6009) to Keqin Hua, the Shanghai Municipal Health Commission (20194Y0085) to Chunbo Li, and the Shanghai “Rising Stars of Medical Talent” Youth Development Program (SHWSRS2020087) to Chunbo Li. Not applicable This study was approved by the Ethical Committee of our hospital (2022-143). The patient provided written informed consent. All data in the study are available from the corresponding author upon reasonable request. Please note: The publisher is not responsible for the content or functionality of any supporting information supplied by the authors. Any queries (other than missing content) should be directed to the corresponding author for the article.

Uniform Manifold Approximation and Projection (UMAP) is a dimensionality-reduction and clustering algorithm designed to transform high-dimensional datasets into optimized low-dimensional embeddings. While UMAP is widely used for efficient and reliable dimensionality reduction, there is a growing demand for improved scalability across heterogeneous datasets. A UMAP implementation in Haskell would boast several advantages over other languages due to Haskell's functional programming properties such as lazy evaluation and static typing, which can provide improved concurrency, parallelization, and scalability, resulting in a more user-friendly experience while enhancing reliability. Despite this, no distributed version of UMAP implemented in Haskell is currently available to the public. To address this gap, this project developed a UMAP implementation that processes fuzzy simplicial sets concurrently, specifically focusing on the parallelization of the K-Nearest-Neighbors (KNN) algorithm, a core component of UMAP. This allows for the distribution of the UMAP algorithm across multiple GPUs as each GPU only keeps a portion of the dataset in VRAM. To accomplish this, a purely functional UMAP was implemented using Haskell (GHC2021) and wrapped in a concurrency monad using the parallel programming library. Typed parsing of datasets was found to be compatible with a functional UMAP implementation, and the fuzzy simplicial sets in UMAP allowed for concurrent processing and the implementation of a scalable UMAP.

Developing a Continuous Severity Scale for Macular Telangiectasia Type 2 Using Deep Learning and Implications for Disease Grading

Non-invasive load monitoring (NILM) deduces changes in energy consumption patterns and operational statuses of electrical equipment from power signals in the feed line. With the emergence of fine-grained power load distribution, the importance of utilizing this technology for implementing demand-side energy management in smart grid development has become increasingly prominent. To address the issue of low load identification accuracy stemming from complex and diverse load types, this paper introduces a NILM method based on uniform manifold approximation and projection (UMAP) reduction and enhanced density-based spatial clustering of applications with noise (DBSCAN). Firstly, this paper combines the characteristics of user load under transient and steady-state conditions and selects data with significant differences to construct a load-characteristic database. Additionally, UMAP is employed to reduce the dimensionality of high-dimensional load features and rebuild a load feature database. Subsequently, DBSCAN is utilized to categorize typical user loads, followed by a correlation analysis with the load-characteristic database to determine the types or classes of loads that involve switching actions. Finally, this paper simulates and analyzes the proposed method using the electricity consumption data of industrial users from the CER–Electricity–Data dataset. It identifies the electricity load data commonly utilized by users in a specific area of Zhejiang Province in China. The experimental results indicate that the accuracy of the proposed non-invasive load identification method reaches 95%. Compared to the wavelet transform, decision tree, and backpropagation network methods, the improvement is approximately 5%.

In situ mineralogical and chemical analyses of rock samples using a space-prototype laser ablation ionization mass spectrometer along with unsupervised machine learning are powerful tools for the study of surface samples on planetary bodies. This potential is demonstrated through the examination of a thin section of a terrestrial rock sample in the laboratory. Autonomous isolation of mineral phases within the acquired mass spectrometric data is achieved with two dimensionality reduction techniques: uniform manifold approximation and projection (UMAP) and density-preserving variation of UMAP (densMAP), and the density-based clustering algorithm Hierarchical Density-Based Spatial Clustering of Applications with Noise (HDBSCAN). Both densMAP and UMAP yield comparable outcomes, successfully isolating the major mineral phases fluorapatite, calcite, and forsterite in the studied rock sample. Notably, densMAP reveals additional insights into the composition of the sample through outlier detection, uncovering signals from the trace minerals pyrite, rutile, baddeleyite, and uranothorianite. Through a grid search, the stability of the methods over a broad model parameter space is confirmed, revealing a correlation between the level of data preprocessing and the resulting clustering quality. Consequently, these methods represent effective strategies for data reduction, highlighting their potential application on board spacecraft to obtain direct and quantitative information on the chemical composition and mineralogy of planetary surfaces and to optimize mission returns through the unsupervised selection of valuable data.

Machine learning uncovers prognostically distinct myeloma cast nephropathy phenotypes not captured by standard risk stratification systems

ABSTRACT: This study explores a rapid and precise method for micromechanical characterization and mapping of heterogeneous rocks, utilizing high-speed nanoindentation and mineral volume fractions. Traditional method for determination of mechanical properties of the different phases of a heterogeneous material require the combination of nanoindentation data and chemical analysis of the material. However, this results in increasing the cost, time, and complexity of the process. Hence, the proposed study explores different data mining techniques such as Uniform Manifold Approximation and Projection (UMAP) with k-means clustering, Dirichlet Process Mixture Model (DPMM) clustering, and Density-Based Spatial Clustering of Applications with Noise (DBSCAN), utilizing high speed nanoindentation data for an efficient and accurate evaluation of the micromechanical properties of heterogeneous shale rock. Comparison of the three techniques deduced that UMAP with k-means clustering technique provides appropriate micromechanical characterization and mapping results with a weighted error of about 13.40%. Even DPMM and DBSCAN performed reasonably well with slightly high weighted errors, therefore they can be adopted as a secondary clustering technique for validation of other clustering technique results. The results demonstrate the potential and efficiency of high-speed nanoindentation test in conjunction with data analytics for characterization and mapping of micromechanical properties of heterogeneous material.

BACKGROUND AND AIMS End-stage renal disease (ESRD) is essentially a chronic inflammatory state and, consequently, it engenders detrimental effects on the immune system. Due to the specific uremic environment, multiple phenotypic lymphocyte alterations are described, similar but not identical to ageing. We aimed to evaluate these changes through simple unsupervised dimensionality reduction algorithms in order to reveal unique phenotypical lymphocyte patterns in ESRD patients. METHOD A wide panel of senescent and exhaustion-related lymphocyte markers, including CD45RA, CCR7, CD31, CD28, CD57, and PD1 on T cells and CD27 and IgD on B cells, was assessed by flow cytometry in 30 ESRD patients and 20 healthy controls of similar age, sex and ethnicity. The resulting immunological multidimensional phenotype was projected at lower dimensions using two algorithms: principal component analysis (PCA) and Uniform Manifold Approximation and Projection (UMAP). RESULTS The plane defined by the two first eigenvectors of PCA showed two clusters, patients and controls, with PCA variable loadings of non-senescent markers pointing towards the controls' centroid. Indeed, ‘naïve’ lymphocytes were reduced in ESRD patients compared with controls [CD4+CD45RA+CCR7+200(150–328) versus 426(260–585) cells/μL, respectively; P = .001] and [CD19+IgD+CD27–54(26–85) versus 130(83–262) cells/μL, respectively; P &amp;lt; .0001]. Also, PCA projections of the multidimensional ESRD immune phenotype suggested a more senescent phenotype in haemodialysis compared with haemodiafiltration treated patients. Finally, clustering based on UMAP projections revealed three distinct patients groups (UMAP 1–3), exhibiting gradual changes for naive, senescent and exhausted lymphocyte markers. Out of these, the UMAP 1 group was characterized by a predominance of CD8 senescent T cell subsets, in contrast, to groups UMAP 2 and 3, which showed a gradual decrease of all T cell subsets, in comparison to healthy controls. These groups were found to differ in the type of dialyzer used, with polysulfone or derivatives alone more often prescribed in UMAP 2–3 patients (13 out of 17) compared with 4 out of 13 in UMAP 1, P = .012. CONCLUSION Simple machine learning algorithms may help to unravel hidden lymphocyte markers and define patterns characteristic of ESRD. Moreover, significant connections of immune changes with dialysis methods and dialyzers were revealed and described.

Dimensionality reduction by UMAP reinforces sample heterogeneity analysis in bulk transcriptomic data.

Single-Cell RNA Sequencing Reveals a Distinct Immune Landscape of Myeloid Cells in Coronary Culprit Plaques Causing Acute Coronary Syndrome.

Objectives: To evaluate the performance of CNN models with feature selection methods like Relief, Uniform Manifold Approximation and Projection (UMAP), and Linear discriminant Analysis (LDA) for forecasting heart diseases. Methods: The present research for heart disease prediction compares the performance of feature selection algorithms like ReliefF, Uniform Manifold Approximation and Projection (UMAP), and Linear discriminant Analysis (LDA) with Convolution Neural networks (CNN) for prediction. The study is conducted in a dataset with 303 records collected from patients with 14 attributes. It is also validated with the publicly available Cleveland dataset (Kaggle). The software environment used for implementation is Jupyter Notebook, which uses Python. The dataset consists of 303 records collected from the patients with 14 attributes. It is validated with the publicly available Cleveland dataset. Findings: The study examines how these feature selection techniques affect CNN's accuracy. According to experimental findings, the CNN-UMAP hybrid model outperforms with an accuracy of 91.88%, precision of 0.89, and recall of 0.85 compared to ReliefF and LDA, UMAP shows up as the most successful feature selection method among the studied techniques when utilized alongside CNNs. Novelty: ReliefF, UMAP, and LDA allow CNN to learn more significant and discriminative features by minimizing the dimensions of the input data while preserving its underlying pattern. Previous studies also attempted to identify the key contributing characteristics to heart disease prediction, but less emphasis was placed on these feature selection methods in determining the effectiveness of the features for heart disease prediction. Keywords: Heart Disease, Feature Selection Methods, Convolutional Neural Network, ReliefF, UMAP and LDA