Advances in AI and machine learning for predictive medicine.

The field of omics, driven by advances in high-throughput sequencing, faces a data explosion. This abundance of data offers unprecedented opportunities for predictive modeling in precision medicine, but also presents formidable challenges in data analysis and interpretation. Traditional machine learning (ML) techniques have been partly successful in generating predictive models for omics analysis but exhibit limitations in handling potential relationships within the data for more accurate prediction. This review explores a revolutionary shift in predictive modeling through the application of deep learning (DL), specifically convolutional neural networks (CNNs). Using transformation methods such as DeepInsight, omics data with independent variables in tabular (table-like, including vector) form can be turned into image-like representations, enabling CNNs to capture latent features effectively. This approach not only enhances predictive power but also leverages transfer learning, reducing computational time, and improving performance. However, integrating CNNs in predictive omics data analysis is not without challenges, including issues related to model interpretability, data heterogeneity, and data size. Addressing these challenges requires a multidisciplinary approach, involving collaborations between ML experts, bioinformatics researchers, biologists, and medical doctors. This review illuminates these complexities and charts a course for future research to unlock the full predictive potential of CNNs in omics data analysis and related fields.

PepCNN deep learning tool for predicting peptide binding residues in proteins using sequence, structural, and language model features.

Protein-peptide interactions play a crucial role in various cellular processes and are implicated in abnormal cellular behaviors leading to diseases such as cancer. Therefore, understanding these interactions is vital for both functional genomics and drug discovery efforts. Despite a significant increase in the availability of protein-peptide complexes, experimental methods for studying these interactions remain laborious, time-consuming, and expensive. Computational methods offer a complementary approach but often fall short in terms of prediction accuracy. To address these challenges, we introduce PepCNN, a deep learning-based prediction model that incorporates structural and sequence-based information from primary protein sequences. By utilizing a combination of half-sphere exposure, position specific scoring matrices from multiple-sequence alignment tool, and embedding from a pre-trained protein language model, PepCNN outperforms state-of-the-art methods in terms of specificity, precision, and AUC. The PepCNN software and datasets are publicly available at https://github.com/abelavit/PepCNN.git .

Time-dependent cell-state selection identifies transiently expressed genes regulating ILC2 activation.

The decision of whether cells are activated or not is controlled through dynamic intracellular molecular networks. However, the low population of cells during the transition state of activation renders the analysis of the transcriptome of this state technically challenging. To address this issue, we have developed the Time-Dependent Cell-State Selection (TDCSS) technique, which employs live-cell imaging of secretion activity to detect an index of the transition state, followed by the simultaneous recovery of indexed cells for subsequent transcriptome analysis. In this study, we used the TDCSS technique to investigate the transition state of group 2 innate lymphoid cells (ILC2s) activation, which is indexed by the onset of interleukin (IL)-13 secretion. The TDCSS approach allowed us to identify time-dependent genes, including transiently induced genes (TIGs). Our findings of IL4 and MIR155HG as TIGs have shown a regulatory function in ILC2s activation.

scDeepInsight: a supervised cell-type identification method for scRNA-seq data with deep learning.

Annotation of cell-types is a critical step in the analysis of single-cell RNA sequencing (scRNA-seq) data that allows the study of heterogeneity across multiple cell populations. Currently, this is most commonly done using unsupervised clustering algorithms, which project single-cell expression data into a lower dimensional space and then cluster cells based on their distances from each other. However, as these methods do not use reference datasets, they can only achieve a rough classification of cell-types, and it is difficult to improve the recognition accuracy further. To effectively solve this issue, we propose a novel supervised annotation method, scDeepInsight. The scDeepInsight method is capable of performing manifold assignments. It is competent in executing data integration through batch normalization, performing supervised training on the reference dataset, doing outlier detection and annotating cell-types on query datasets. Moreover, it can help identify active genes or marker genes related to cell-types. The training of the scDeepInsight model is performed in a unique way. Tabular scRNA-seq data are first converted to corresponding images through the DeepInsight methodology. DeepInsight can create a trainable image transformer to convert non-image RNA data to images by comprehensively comparing interrelationships among multiple genes. Subsequently, the converted images are fed into convolutional neural networks such as EfficientNet-b3. This enables automatic feature extraction to identify the cell-types of scRNA-seq samples. We benchmarked scDeepInsight with six other mainstream cell annotation methods. The average accuracy rate of scDeepInsight reached 87.5%, which is more than 7% higher compared with the state-of-the-art methods.

Topologically associating domain underlies tissue specific expression of long intergenic non-coding RNAs.

Accumulating evidence indicates that long intergenic non-coding RNAs (lincRNAs) show more tissue-specific expression patterns than protein-coding genes (PCGs). However, although lincRNAs are subject to canonical transcriptional regulation like PCGs, the molecular basis for the specificity of their expression patterns remains unclear. Here, using expression data and coordinates of topologically associating domains (TADs) in human tissues, we show that lincRNA loci are significantly enriched in the more internal region of TADs compared to PCGs and that lincRNAs within TADs have higher tissue specificity than those outside TADs. Based on these, we propose an analytical framework to interpret transcriptional status using lincRNA as an indicator. We applied it to hypertrophic cardiomyopathy data and found disease-specific transcriptional regulation: ectopic expression of keratin at the TAD level and derepression of myocyte differentiation-related genes by E2F1 with down-regulation of LINC00881. Our results provide understanding of the function and regulation of lincRNAs according to genomic structure.

The early neutrophil-committed progenitors aberrantly differentiate into immunoregulatory monocytes during emergency myelopoiesis.

Inflammatory stimuli cause a state of emergency myelopoiesis leading to neutrophil-like monocyte expansion. However, their function, the committed precursors, or growth factors remain elusive. In this study we find that Ym1+Ly6Chi monocytes, an immunoregulatory entity of neutrophil-like monocytes, arise from progenitors of neutrophil 1 (proNeu1). Granulocyte-colony stimulating factor (G-CSF) favors the production of neutrophil-like monocytes through previously unknown CD81+CX3CR1lo monocyte precursors. GFI1 promotes the differentiation of proNeu2 from proNeu1 at the cost of producing neutrophil-like monocytes. The human counterpart of neutrophil-like monocytes that also expands in response to G-CSF is found in CD14+CD16- monocyte fraction. The human neutrophil-like monocytes are discriminated from CD14+CD16- classical monocytes by CXCR1 expression and the capacity to suppress T cell proliferation. Collectively, our findings suggest that the aberrant expansion of neutrophil-like monocytes under inflammatory conditions is a process conserved between mouse and human, which may be beneficial for the resolution of inflammation.

DeepInsight-3D architecture for anti-cancer drug response prediction with deep-learning on multi-omics.

Modern oncology offers a wide range of treatments and therefore choosing the best option for particular patient is very important for optimal outcome. Multi-omics profiling in combination with AI-based predictive models have great potential for streamlining these treatment decisions. However, these encouraging developments continue to be hampered by very high dimensionality of the datasets in combination with insufficiently large numbers of annotated samples. Here we proposed a novel deep learning-based method to predict patient-specific anticancer drug response from three types of multi-omics data. The proposed DeepInsight-3D approach relies on structured data-to-image conversion that then allows use of convolutional neural networks, which are particularly robust to high dimensionality of the inputs while retaining capabilities to model highly complex relationships between variables. Of particular note, we demonstrate that in this formalism additional channels of an image can be effectively used to accommodate data from different omics layers while implicitly encoding the connection between them. DeepInsight-3D was able to outperform other state-of-the-art methods applied to this task. The proposed improvements can facilitate the development of better personalized treatment strategies for different cancers in the future.

CNN-Pred: Prediction of single-stranded and double-stranded DNA-binding protein using convolutional neural networks.

DNA-binding proteins play a vital role in biological activity including DNA replication, DNA packing, and DNA reparation. DNA-binding proteins can be classified into single-stranded DNA-binding proteins (SSBs) or double-stranded DNA-binding proteins (DSBs). Determining whether a protein is DSB or SSB helps determine the protein's function. Therefore, many studies have been conducted to accurately identify DSB and SSB in recent years. Despite all the efforts have been made so far, the DSB and SSB prediction performance remains limited. In this study, we propose a new method called CNN-Pred to accurately predict DSB and SSB. To build CNN-Pred, we first extract evolutionary-based features in the form of mono-gram and bi-gram profiles using position specific scoring matrix (PSSM). We then, use 1D-convolutional neural network (CNN) as the classifier to our extracted features. Our results demonstrate that CNN-Pred can enhance the DSB and SSB prediction accuracies by more than 4%, on the independent test compared to previous studies found in the literature. CNN-pred as a standalone tool and all its source codes are publicly available at: https://github.com/MLBC-lab/CNN-Pred.

Targeted deep sequencing analyses of long QT syndrome in a Japanese population.

Long QT syndrome (LQTS) is one of the most common inherited arrhythmias and multiple genes have been reported as causative. Presently, genetic diagnosis for LQTS patients is becoming widespread and contributing to implementation of therapies. However, causative genetic mutations cannot be detected in about 20% of patients. To elucidate additional genetic mutations in LQTS, we performed deep-sequencing of previously reported 15 causative and 85 candidate genes for this disorder in 556 Japanese LQTS patients. We performed in-silico filtering of the sequencing data and found 48 novel variants in 33 genes of 53 cases. These variants were predicted to be damaging to coding proteins or to alter the binding affinity of several transcription factors. Notably, we found that most of the LQTS-related variants in the RYR2 gene were in the large cytoplasmic domain of the N-terminus side. They might be useful for screening of LQTS patients who had no known genetic factors. In addition, when the mechanisms of these variants in the development of LQTS are revealed, it will be useful for early diagnosis, risk stratification, and selection of treatment.

Utility of tissue-specific gene expression scores for gene prioritization in Mendelian diseases.

In genetic testing of Mendelian diseases, it is a bioinformatics challenge to effectively prioritize disease-causing candidate genes listed from massively parallel sequencing. Tissue specificity of the gene expression levels may give a clue because it may reflect tissue-specific disease manifestation. However, considering poor correlations between mRNA and protein expression in some genes, it is not clear whether transcriptomics- or proteomics-based tissue specificity should be used to prioritize candidate genes. Therefore, we compared the efficiency of tissue-specific scores (TS scores) obtained from transcriptome and proteome data in prioritizing candidate genes for whole exome sequencing (WES) analysis of Mendelian disease patients. We show that both Protein and RNA TS scores are useful in prioritizing candidate genes in WES analysis, although diseases like coagulopathies get more benefit from Protein TS score. This study may provide useful evidence in developing new methods to effectively identify novel disease-causing genes.

Splicing QTL analysis focusing on coding sequences reveals mechanisms for disease susceptibility loci.

Splicing quantitative trait loci (sQTLs) are one of the major causal mechanisms in genome-wide association study (GWAS) loci, but their role in disease pathogenesis is poorly understood. One reason is the complexity of alternative splicing events producing many unknown isoforms. Here, we propose two approaches, namely integration and selection, for this complexity by focusing on protein-structure of isoforms. First, we integrate isoforms with the same coding sequence (CDS) and identify 369-601 integrated-isoform ratio QTLs (i2-rQTLs), which altered protein-structure, in six immune subsets. Second, we select CDS incomplete isoforms annotated in GENCODE and identify 175-337 isoform-ratio QTL (i-rQTL). By comprehensive long-read capture RNA-sequencing among these incomplete isoforms, we reveal 29 full-length isoforms with unannotated CDSs associated with GWAS traits. Furthermore, we show that disease-causal sQTL genes can be identified by evaluating their trans-eQTL effects. Our approaches highlight the understudied role of protein-altering sQTLs and are broadly applicable to other tissues and diseases.

Revisiting the definition of glioma recurrence based on a phylogenetic investigation of primary and re-emerging tumor samples: a case report.

A recurrent tumor is defined as a re-emerging subclone originating from an ancestorial clone of the primary neoplasm. Hence, it should be distinguished from de novo tumor emerging from other clones. Herein, we describe an exceptional case in which the locally re-emerging glioma did not share genetic alterations of the primary tumor. While the initial tumor harbored mutations in IDH1 and TERT genes as well as 1p/19q codeletion, the re-emerging tumor did not present any of these genetic abnormalities. Variant calling for tumor samples using whole-genome sequencing revealed that 1696 mutations within the primary tumor faded in the re-emerging tumor, and that 4591 mutations were newly detected in the re-emerging tumor. These results suggested that the initial and re-emerging tumors did not share same clonal origins, although the second tumor appeared adjacent to the old surgical cavity 5 years after the initial surgery. We finally speculated that the re-emerging tumor could be a "de novo glioma" or "radiation-induced glioblastoma following treatment of a diffuse glioma." This case highlights the importance of molecular re-evaluation of clinically diagnosed "recurrent" glioma lesions.

Principal component analysis of early immune cell dynamics during pembrolizumab treatment of advanced urothelial carcinoma.

Immune checkpoint inhibitors have been approved as second-line therapy for patients with advanced urothelial carcinoma (UC). However, which patients will obtain clinical benefit remains to be determined. To identify predictive biomarkers for the pembrolizumab (PEM) response early during treatment, the present study investigated 31 patients with chemotherapy-resistant recurrent or metastatic UC who received 200 mg PEM intravenously every 3 weeks. Blood was taken just before the first dose and again before the second dose, and the peripheral blood mononuclear cells of all 31 pairs of blood samples were immune phenotyped by flow cytometry. Data were assessed by principal component analysis (PCA), correlation analysis and Cox proportional hazards modeling in order to comprehensively determine the effects of PEM on peripheral mononuclear immune cells. Absolute counts of CD45RA+CD27-CCR7- terminally differentiated CD8+ T cells and KLRG1+CD57+ senescent CD8+ T cells were significantly increased after PEM administration (P=0.042 and P=0.043, respectively). Senescent and exhausted CD4+ and CD8+ T cell dynamics were strongly associated with each other. By contrast, counts of monocytic myeloid-derived suppressor cells (mMDSCs) were not associated with other immune cell phenotypes. The results of PCA and non-hierarchical clustering of patients suggested that excessive T cell senescence and differentiation early during treatment were not necessarily associated with a survival benefit. However, decreased mMDSC counts after PEM were associated with improved overall survival. In conclusion, early on-treatment peripheral T cell status was associated with response to PEM; however, it was not associated with clinical benefit. By contrast, decreased peripheral mMDSC counts did predict improved overall survival.

Association between high immune activity and worse prognosis in uveal melanoma and low-grade glioma in TCGA transcriptomic data.

BACKGROUND: Immune status in the tumor microenvironment is an important determinant of cancer progression and patient prognosis. Although a higher immune activity is often associated with a better prognosis, this trend is not absolute and differs across cancer types. We aimed to give insights into why some cancers do not show better survival despite higher immunity by assessing the relationship between different biological factors, including cytotoxicity, and patient prognosis in various cancer types using RNA-seq data collected by The Cancer Genome Atlas. RESULTS: Results showed that a higher immune activity was associated with worse overall survival in patients with uveal melanoma and low-grade glioma, which are cancers of immune-privileged sites. In these cancers, epithelial or endothelial mesenchymal transition and inflammatory state as well as immune activation had a notable negative correlation with patient survival. Further analysis using additional single-cell data of uveal melanoma and glioma revealed that epithelial or endothelial mesenchymal transition was mainly induced in retinal pigment cells or endothelial cells that comprise the blood-retinal and blood-brain barriers, which are unique structures of the eye and central nervous system, respectively. Inflammation was mainly promoted by macrophages, and their infiltration increased significantly in response to immune activation. Furthermore, we found the expression of inflammatory chemokines, particularly CCL5, was strongly correlated with immune activity and associated with poor survival, particularly in these cancers, suggesting that these inflammatory mediators are potential molecular targets for therapeutics. CONCLUSIONS: In uveal melanoma and low-grade glioma, inflammation from macrophages and epithelial or endothelial mesenchymal transition are particularly associated with a poor prognosis. This implies that they loosen the structures of the blood barrier and impair homeostasis and further recruit immune cells, which could result in a feedback loop of additional inflammatory effects leading to runaway conditions.

Unique characteristics of tertiary lymphoid structures in kidney clear cell carcinoma: prognostic outcome and comparison with bladder cancer.

BACKGROUND: The aims of this study were (1) to clarify the impact of tertiary lymphoid structure (TLS) status on the outcome and immunogenomic profile of human clear cell renal cell carcinoma (ccRCC) and (2) to determine phenotypic differences in TLSs between different types of genitourinary cancer, that is, urinary ccRCC and bladder cancer. METHODS: We performed a quantitative immunohistological analysis of ccRCC tissue microarrays and conducted integrated genome mutation analysis by next-generation sequencing and methylation array analysis. Since the tumor immune microenvironment of ccRCC often differs from that of other cancer types, we analyzed the phenotypic differences in TLSs between ccRCC and in-house bladder cancer specimens. RESULTS: Varying distribution patterns of TLSs were observed throughout ccRCC tumors, revealing that the presence of TLSs was related to poor prognosis. An analysis of genomic alterations based on TLS status in ccRCC revealed that alterations in the PI3K-mTOR pathway were highly prevalent in TLS-positive tumors. DNA methylation profiling also revealed distinct differences in methylation signatures among ccRCC samples with different TLS statuses. However, the TLS characteristics of ccRCC and bladder cancer markedly differed: TLSs had the exact opposite prognostic impact on bladder cancer as on ccRCC. The maturity and spatial distribution of TLSs were significantly different between the two cancer types; TLSs were more mature with follicle-like germinal center organization and likely to be observed inside the tumor in bladder cancer. Labeling for CD8, FOXP3, PD-1, and PD-L1 showed marked differences in the diversity of the immune microenvironment surrounding TLSs. The proportions of CD8-, FOXP3-, and PD-L1-positive cells were significantly higher in TLSs in bladder cancer than in TLSs in ccRCC; rather the proportion of PD-1-positive cells was significantly higher in TLSs in ccRCC than in TLSs in bladder cancer. CONCLUSION: The immunobiology of ccRCC is unique, and various cancerous phenomena conflict with that seen in other cancer types; therefore, comparing the TLS characteristics between ccRCC and bladder cancer may help reveal differences in the prognostic impact, maturity and spatial distribution of TLSs and in the immune environment surrounding TLSs between the two cancers.

Immune subtypes and neoantigen-related immune evasion in advanced colorectal cancer.

Elimination of cancerous cells by the immune system is an important mechanism of protection from cancer, however, its effectiveness can be reduced owing to development of resistance and evasion. To understand the systemic immune response in advanced untreated primary colorectal cancer, we analyze immune subtypes and immune evasion via neoantigen-related mechanisms. We identify a distinctive cancer subtype characterized by immune evasion and very poor overall survival. This subtype has less clonal highly expressed neoantigens and high chromosomal instability, resulting in adaptive immune resistance mediated by the immune checkpoint molecules and neoantigen presentation disorders. We also observe that neoantigen depletion caused by immunoediting and high clonal neoantigen load are correlated with a good overall survival. Our results indicate that the status of the tumor microenvironment and neoantigen composition are promising new prognostic biomarkers with potential relevance for treatment plan decisions in advanced CRC.

Four pedigrees with aminoacyl-tRNA synthetase abnormalities.

Aminoacyl tRNA synthetases (ARSs) are highly conserved enzymes that link amino acids to their cognate tRNAs. Thirty-seven ARSs are known and their deficiencies cause various genetic disorders. Variants in some ARSs are associated with the autosomal dominant inherited form of axonal neuropathy, including Charcot-Marie-Tooth (CMT) disease. Variants of genes encoding ARSs often cause disorders in an autosomal recessive fashion. The clinical features of cytosolic ARS deficiencies are more variable, including systemic features. Deficiencies of ARSs localized in the mitochondria are often associated with neurological disorders including Leigh and early-onset epileptic syndromes. Whole exome sequencing (WES) is an efficient way to identify the genes causing various symptoms in patients. We identified 4 pedigrees with novel compound heterozygous variants in ARS genes (WARS1, MARS1, AARS2, and PARS2) by WES. Some unique manifestations were noted. The number of patients with ARSs has been increasing since the application of WES. Our findings broaden the known genetic and clinical spectrum associated with ARS variants.

Hemorrhagic shock and encephalopathy syndrome in a patient with a de novo heterozygous variant in KIF1A.

INTRODUCTION: KIF1A, a gene that encodes a neuron-specific motor protein, plays important roles in cargo transport along neurites. Variants in KIF1A have been described in three different disorders, and neurodegeneration and spasticity with or without cerebellar atrophy or cortical visual impairment syndrome (NESCAVS) is the severest phenotype. CASE REPORT: A 3-year-old girl was born at term with a birth weight of 2590 g. At five months of age, she visited our hospital due to developmental delay. An EEG showed multiple epileptic discharge, and a nerve conduction study showed severe axonopathy of both motor and sensory nerves. We performed exome sequencing and identified a de novo heterozygous missense variant in KIF1A (NM_001244008.1: c. 757G > A, p.E253K). At six months of age, she developed acute encephalopathy, multiple organ failure and disseminated intravascular coagulation, necessitating intensive care. Her brain CT showed severe brain edema, followed by profound brain atrophy. We diagnosed hemorrhagic shock and encephalopathy syndrome (HSES) according to the clinico-radiological features. Currently, she is bed-ridden, and requires gastrostomy because of dysphagia. CONCLUSION: The clinical course of our case confirmed that p.E253K is associated with severe neurological features. Severe KIF1A deficiency could cause thermoregulatory dysfunction and may increase the risk of acute encephalopathy including HSES.

Landscape of prognostic signatures and immunogenomics of the AXL/GAS6 axis in renal cell carcinoma.

BACKGROUND: Cabozantinib is an oral tyrosine kinase inhibitor in renal cell carcinoma (RCC), whose targets include oncogenic AXL and unique ligand GAS6. Critical gaps in basic knowledge need to be addressed to devise an exclusive biomarker and candidate when targeting the AXL/GAS6 axis. METHODS: To clarify the effects of the AXL/GAS6 axis on RCC, we herein performed a large-scale immunogenomic analysis and single-cell counts including various metastatic organs and histological subtypes of RCC. We further applied genome-wide mutation analyses and methylation arrays. RESULTS: Varying patterns of AXL and GAS6 expression were observed throughout primary RCC tumours and metastases. Scoring individual AXL/GAS6 levels in the tumour centre and invasive margin, namely, the AXL/GAS6 score, showed a good ability to predict the prognosis of clear cell RCC. Metastasis- and histological subtype-specific differences in the AXL/GAS6 score existed since lung metastasis and the papillary subtype were weakly related to the AXL/GAS6 axis. Cell-by-cell immunohistological assessments clarified an immunosuppressive environment in tumours with high AXL/GAS6 scores. Genomic alterations in the PI3K-mTOR pathway and DNA methylation profiling revealed distinct differences with the AXL/GAS6 score in ccRCC. CONCLUSION: The AXL/GAS6 scoring system could predict the outcome of prognosis and work as a robust biomarker for the immunogenomic state in RCC.

Profiling the inhibitory receptors LAG-3, TIM-3, and TIGIT in renal cell carcinoma reveals malignancy.

A cutting edge therapy for future immuno-oncology is targeting a new series of inhibitory receptors (IRs): LAG-3, TIM-3, and TIGIT. Both immunogenomic analyses and diagnostic platforms to distinguish candidates and predict good responders to these IR-related agents are vital in clinical pathology. By applying an automated single-cell count for immunolabelled LAG-3, TIM-3, and TIGIT, we reveal that individual IR levels with exclusive domination in each tumour can serve as valid biomarkers for profiling human renal cell carcinoma (RCC). We uncover the immunogenomic landscape associated with individual IR levels in human RCC tumours with metastases in various organs and histological subtypes. We then externally validate our results and devise a workflow with optimal biomarker cut-offs for discriminating the LAG-3, TIM-3, and TIGIT tumour profiles. The discrimination of LAG-3, TIM-3, and TIGIT profiles in tumours may have a broad impact on investigations of immunotherapy responses after targeting a new series of IRs.