Large language model produces high accurate diagnosis of cancer from end-motif profiles of cell-free DNA.

Instruction-tuned large language models (LLMs) demonstrate exceptional ability to align with human intentions. We present an LLM-based model-instruction-tuned LLM for assessment of cancer (iLLMAC)-that can detect cancer using cell-free deoxyribonucleic acid (cfDNA) end-motif profiles. Developed on plasma cfDNA sequencing data from 1135 cancer patients and 1106 controls across three datasets, iLLMAC achieved area under the receiver operating curve (AUROC) of 0.866 [95% confidence interval (CI), 0.773-0.959] for cancer diagnosis and 0.924 (95% CI, 0.841-1.0) for hepatocellular carcinoma (HCC) detection using 16 end-motifs. Performance increased with more motifs, reaching 0.886 (95% CI, 0.794-0.977) and 0.956 (95% CI, 0.89-1.0) for cancer diagnosis and HCC detection, respectively, with 64 end-motifs. On an external-testing set, iLLMAC achieved AUROC of 0.912 (95% CI, 0.849-0.976) for cancer diagnosis and 0.938 (95% CI, 0.885-0.992) for HCC detection with 64 end-motifs, significantly outperforming benchmarked methods. Furthermore, iLLMAC achieved high classification performance on datasets with bisulfite and 5-hydroxymethylcytosine sequencing. Our study highlights the effectiveness of LLM-based instruction-tuning for cfDNA-based cancer detection.

Language model enables end-to-end accurate detection of cancer from cell-free DNA.

We present a language model Affordable Cancer Interception and Diagnostics (ACID) that can achieve high classification performance in the diagnosis of cancer exclusively from using raw cfDNA sequencing reads. We formulate ACID as an autoregressive language model. ACID is pretrained with language sentences that are obtained from concatenation of raw sequencing reads and diagnostic labels. We benchmark ACID against three methods. On testing set subjected to whole-genome sequencing, ACID significantly outperforms the best benchmarked method in diagnosis of cancer [Area Under the Receiver Operating Curve (AUROC), 0.924 versus 0.853; P < 0.001] and detection of hepatocellular carcinoma (AUROC, 0.981 versus 0.917; P < 0.001). ACID can achieve high accuracy with just 10 000 reads per sample. Meanwhile, ACID achieves the best performance on testing sets that were subjected to bisulfite sequencing compared with benchmarked methods. In summary, we present an affordable, simple yet efficient end-to-end paradigm for cancer detection using raw cfDNA sequencing reads.

Scalable batch-correction approach for integrating large-scale single-cell transcriptomes.

Integration of accumulative large-scale single-cell transcriptomes requires scalable batch-correction approaches. Here we propose Fugue, a simple and efficient batch-correction method that is scalable for integrating super large-scale single-cell transcriptomes from diverse sources. The core idea of the method is to encode batch information as trainable parameters and add it to single-cell expression profile; subsequently, a contrastive learning approach is used to learn feature representation of the additive expression profile. We demonstrate the scalability of Fugue by integrating all single cells obtained from the Human Cell Atlas. We benchmark Fugue against current state-of-the-art methods and show that Fugue consistently achieves improved performance in terms of data alignment and clustering preservation. Our study will facilitate the integration of single-cell transcriptomes at increasingly large scale.

A universal approach for integrating super large-scale single-cell transcriptomes by exploring gene rankings.

Advancement in single-cell RNA sequencing leads to exponential accumulation of single-cell expression data. However, there is still lack of tools that could integrate these unlimited accumulations of single-cell expression data. Here, we presented a universal approach iSEEEK for integrating super large-scale single-cell expression via exploring expression rankings of top-expressing genes. We developed iSEEEK with 11.9 million single cells. We demonstrated the efficiency of iSEEEK with canonical single-cell downstream tasks on five heterogenous datasets encompassing human and mouse samples. iSEEEK achieved good clustering performance benchmarked against well-annotated cell labels. In addition, iSEEEK could transfer its knowledge learned from large-scale expression data on new dataset that was not involved in its development. iSEEEK enables identification of gene-gene interaction networks that are characteristic of specific cell types. Our study presents a simple and yet effective method to integrate super large-scale single-cell transcriptomes and would facilitate translational single-cell research from bench to bedside.

Development of an AI system for accurately diagnose hepatocellular carcinoma from computed tomography imaging data.

BACKGROUND AND AIMS: Computed tomography (CT) scan is frequently used to detect hepatocellular carcinoma (HCC) in routine clinical practice. The aim of this study is to develop a deep-learning AI system to improve the diagnostic accuracy of HCC by analysing liver CT imaging data. METHODS: We developed a deep-learning AI system by training on CT images from 7512 patients at Henan Provincial Peoples' Hospital. Its performance was validated on one internal test set (Henan Provincial Peoples' Hospital, n = 385) and one external test set (Henan Provincial Cancer Hospital, n = 556). The area under the receiver-operating characteristic curve (AUROC) was used as the primary classification metric. Accuracy, sensitivity, specificity, precision, negative predictive value and F1 metric were used to measure the performance of AI systems and radiologists. RESULTS: AI system achieved high performance in identifying HCC patients, with AUROC of 0.887 (95% CI 0.855-0.919) on the internal test set and 0.883 (95% CI 0.855-0.911) on the external test set. For internal test set, accuracy was 81.0% (76.8-84.8%), sensitivity was 78.4% (72.4-83.7%), specificity was 84.4% (78.0-89.6%) and F1 (harmonic average of precision and recall rate) was 0.824. For external test set, accuracy was 81.3% (77.8-84.5%), sensitivity was 89.4% (85.0-92.8%), specificity was 74.0% (68.5-78.9%) and F1 was 0.819. Compared with radiologists, AI system achieved comparable accuracy and F1 metric on internal test set (0.853 versus 0.818, P = 0.107; 0.863 vs. 0.824, P = 0.082) and external test set (0.805 vs. 0.793, P = 0.663; 0.810 vs. 0.814, P = 0.866). The predicted HCC risk scores by AI system in HCC patients with multiple tumours and high fibrosis stage were higher than those with solitary tumour and low fibrosis stage (tumour number: 0.197 vs. 0.138, P = 0.006; fibrosis stage: 0.183 vs. 0.127, P < 0.001). Radiologists' review showed that the accuracy of saliency heatmaps predicted by algorithms was 92.1% (95% CI: 89.2-95.0%). CONCLUSIONS: AI system achieved high performance in the detection of HCC compared with a group of specialised radiologists. Further investigation by prospective clinical trials was necessitated to verify this model.

Highly Oxygenated Organic Nitrates Formed from NO3 Radical-Initiated Oxidation of ß-Pinene.

The reactions of biogenic volatile organic compounds (BVOC) with the nitrate radicals (NO3) are major night-time sources of organic nitrates and secondary organic aerosols (SOA) in regions influenced by BVOC and anthropogenic emissions. In this study, the formation of gas-phase highly oxygenated organic molecules-organic nitrates (HOM-ON) from NO3-initiated oxidation of a representative monoterpene, ß-pinene, was investigated in the SAPHIR chamber (Simulation of Atmosphere PHotochemistry In a large Reaction chamber). Six monomer (C = 7-10, N = 1-2, O = 6-16) and five accretion product (C = 17-20, N = 2-4, O = 9-22) families were identified and further classified into first- or second-generation products based on their temporal behavior. The time lag observed in the peak concentrations between peroxy radicals containing odd and even number of oxygen atoms, as well as between radicals and their corresponding termination products, provided constraints on the HOM-ON formation mechanism. The HOM-ON formation can be explained by unimolecular or bimolecular reactions of peroxy radicals. A dominant portion of carbonylnitrates in HOM-ON was detected, highlighting the significance of unimolecular termination reactions by intramolecular H-shift for the formation of HOM-ON. A mean molar yield of HOM-ON was estimated to be 4.8% (-2.6%/+5.6%), suggesting significant HOM-ON contributions to the SOA formation.

Elevated Concentrations of Lead in Particulate Matter on the Neighborhood-Scale in Delhi, India As Determined by Single Particle Analysis.

High mass concentrations of atmospheric lead particles are frequently observed in the Delhi, India metropolitan area, although the sources of lead particles are poorly understood. In this study, particles sampled across Delhi (August - December 2008) were analyzed by computer-controlled scanning electron microscopy with energy dispersive X-ray spectroscopy (CCSEM-EDX) to improve our understanding of the spatial and physicochemical variability of lead-rich particles (>90% lead). The mean mass concentration of lead-rich particles smaller than 10 µm (PM10) was 0.7 µg/m(3) (1.5 µg/m(3) std. dev.) with high variability (range: 0-6.2 µg/m(3)). Four samples (16% of 25 samples) with PM10 lead-rich particle concentrations >1.4 µg/m(3) were defined as lead events and studied further. The temporal characteristics, heterogeneous spatial distribution, and wind patterns of events, excluded regional monsoon conditions or common anthropogenic sources from being the major causes of the lead events. Individual particle composition, size, and morphology analysis indicate informal recycling operations of used lead-acid batteries as the likely source of the lead events. This source is not typically included in emission inventories, and the observed isolated hotspots with high lead concentrations could represent an elevated exposure risk in certain neighborhoods of Delhi.

Development of a deep learning model for cancer diagnosis by inspecting cell-free DNA end-motifs.

Accurate discrimination between patients with and without cancer from cfDNA is crucial for early cancer diagnosis. Herein, we develop and validate a deep-learning-based model entitled end-motif inspection via transformer (EMIT) for discriminating individuals with and without cancer by learning feature representations from cfDNA end-motifs. EMIT is a self-supervised learning approach that models rankings of cfDNA end-motifs. We include 4606 samples subjected to different types of cfDNA sequencing to develop EIMIT, and subsequently evaluate classification performance of linear projections of EMIT on six datasets and an additional inhouse testing set encopassing whole-genome, whole-genome bisulfite and 5-hydroxymethylcytosine sequencing. The linear projection of representations from EMIT achieved area under the receiver operating curve (AUROC) values ranged from 0.895 (0.835-0.955) to 0.996 (0.994-0.997) across these six datasets, outperforming its baseline by significant margins. Additionally, we showed that linear projection of EMIT representations can achieve an AUROC of 0.962 (0.914-1.0) in identification of lung cancer on an independent testing set subjected to whole-exome sequencing. The findings of this study indicate that a transformer-based deep learning model can learn cancer-discrimative representations from cfDNA end-motifs. The representations of this deep learning model can be exploited for discriminating patients with and without cancer.

Transformer-based representation learning and multiple-instance learning for cancer diagnosis exclusively from raw sequencing fragments of bisulfite-treated plasma cell-free DNA.

Early cancer diagnosis from bisulfite-treated cell-free DNA (cfDNA) fragments requires tedious data analytical procedures. Here, we present a deep-learning-based approach for early cancer interception and diagnosis (DECIDIA) that can achieve accurate cancer diagnosis exclusively from bisulfite-treated cfDNA sequencing fragments. DECIDIA relies on transformer-based representation learning of DNA fragments and weakly supervised multiple-instance learning for classification. We systematically evaluate the performance of DECIDIA for cancer diagnosis and cancer type prediction on a curated dataset of 5389 samples that consist of colorectal cancer (CRC; n = 1574), hepatocellular cell carcinoma (HCC; n = 1181), lung cancer (n = 654), and non-cancer control (n = 1980). DECIDIA achieved an area under the receiver operating curve (AUROC) of 0.980 (95% CI, 0.976-0.984) in 10-fold cross-validation settings on the CRC dataset by differentiating cancer patients from cancer-free controls, outperforming benchmarked methods that are based on methylation intensities. Noticeably, DECIDIA achieved an AUROC of 0.910 (95% CI, 0.896-0.924) on the externally independent HCC testing set in distinguishing HCC patients from cancer-free controls, although there was no HCC data used in model development. In the settings of cancer-type classification, we observed that DECIDIA achieved a micro-average AUROC of 0.963 (95% CI, 0.960-0.966) and an overall accuracy of 82.8% (95% CI, 81.8-83.9). In addition, we distilled four sequence signatures from the raw sequencing reads that exhibited differential patterns in cancer versus control and among different cancer types. Our approach represents a new paradigm towards eliminating the tedious data analytical procedures for liquid biopsy that uses bisulfite-treated cfDNA methylome.

Macrophage-coated tumor cluster aggravates hepatoma invasion and immunotherapy resistance via generating local immune deprivation.

Immune checkpoint inhibitors (ICIs) represent a promising treatment for hepatocellular carcinoma (HCC) due to their capacity for abundant lymphocyte infiltration. However, some patients with HCC respond poorly to ICI therapy due to the presence of various immunosuppressive factors in the tumor microenvironment. Our research reveals that a macrophage-coated tumor cluster (MCTC) signifies a unique spatial structural organization in HCC correlating with diminished recurrence-free survival and overall survival in a total of 572 HCC cases from 3 internal cohorts and 2 independent external validation cohorts. Mechanistically, tumor-derived macrophage-associated lectin Mac-2 binding protein (M2BP) induces MCTC formation and traps immunocompetent cells at the edge of MCTCs to induce intratumoral cytotoxic T cell exclusion and local immune deprivation. Blocking M2BP with a Mac-2 antagonist might provide an effective approach to prevent MCTC formation, enhance T cell infiltration, and thereby improve the efficacy of ICI therapy in HCC.

An efficient context-aware approach for whole-slide image classification.

Computational pathology for gigapixel whole-slide images (WSIs) at slide level is helpful in disease diagnosis and remains challenging. We propose a context-aware approach termed WSI inspection via transformer (WIT) for slide-level classification via holistically modeling dependencies among patches on WSI. WIT automatically learns feature representation of WSI by aggregating features of all image patches. We evaluate classification performance of WIT and state-of-the-art baseline method. WIT achieved an accuracy of 82.1% (95% CI, 80.7%-83.3%) in the detection of 32 cancer types on the TCGA dataset, 0.918 (0.910-0.925) in diagnosis of cancer on the CPTAC dataset, and 0.882 (0.87-0.890) in the diagnosis of prostate cancer from needle biopsy slide, outperforming the baseline by 31.6%, 5.4%, and 9.3%, respectively. WIT can pinpoint the WSI regions that are most influential for its decision. WIT represents a new paradigm for computational pathology, facilitating the development of digital pathology tools.

Generative pretraining from large-scale transcriptomes for single-cell deciphering.

Exponential accumulation of single-cell transcriptomes poses great challenge for efficient assimilation. Here, we present an approach entitled generative pretraining from transcriptomes (tGPT) for learning feature representation of transcriptomes. tGPT is conceptually simple in that it autoregressive models the ranking of a gene in the context of its preceding neighbors. We developed tGPT with 22.3 million single-cell transcriptomes and used four single-cell datasets to evalutate its performance on single-cell analysis tasks. In addition, we examine its applications on bulk tissues. The single-cell clusters and cell lineage trajectories derived from tGPT are highly aligned with known cell labels and states. The feature patterns of tumor bulk tissues learned by tGPT are associated with a wide range of genomic alteration events, prognosis, and treatment outcome of immunotherapy. tGPT represents a new analytical paradigm for integrating and deciphering massive amounts of transcriptome data and it will facilitate the interpretation and clinical translation of single-cell transcriptomes.

A Multiparametric Fusion Radiomics Signature Based on Contrast-Enhanced MRI for Predicting Early Recurrence of Hepatocellular Carcinoma.

Objectives: The postoperative early recurrence (ER) rate of hepatocellular carcinoma (HCC) is 50%, and no highly reliable predictive tool has been developed yet. The aim of this study was to develop and validate a predictive model with radiomics analysis based on multiparametric magnetic resonance (MR) images to predict early recurrence of HCC. Methods: In total, 302 patients (training dataset: n = 211; validation dataset: n = 91) with pathologically confirmed HCC who underwent preoperative MR imaging were enrolled in this study. Three-dimensional regions of interest of the entire lesion were accessed by manually drawing along the tumor margins on the multiple sequences of MR images. Least absolute shrinkage and selection operator Cox regression were then applied to select ER-related radiomics features and construct radiomics signatures. Univariate analysis and multivariate Cox regression analysis were used to identify the significant clinico-radiological factors and establish a clinico-radiological model. A predictive model of ER incorporating the fusion radiomics signature and clinico-radiological risk factors was constructed. The diagnostic performance and clinical utility of this model were measured by receiver-operating characteristic (ROC), calibration curve, and decision curve analyses. Results: The fusion radiomics signature consisting of 6 radiomics features achieved good prediction performance (training dataset: AUC = 0.85, validation dataset: AUC = 0.79). The predictive model of ER integrating clinico-radiological risk factors and the fusion radiomics signature improved the prediction efficacy with AUCs of 0.91 and 0.87 in the training and validation datasets, respectively. Furthermore, the nomogram and ER risk stratification system based on the predictive model demonstrated encouraging predictions of the individualized risk of ER and gave three risk groups with low, intermediate, or high risk of ER. Conclusions: The proposed predictive model incorporating clinico-radiological factors and the fusion radiomics signature derived from multiparametric MR images may be an effective tool for the individualized prediction of postoperative ER in patients with HCC.

Unexpected significance of a minor reaction pathway in daytime formation of biogenic highly oxygenated organic compounds.

Secondary organic aerosol (SOA), formed by oxidation of volatile organic compounds, substantially influence air quality and climate. Highly oxygenated organic molecules (HOMs), particularly those formed from biogenic monoterpenes, contribute a large fraction of SOA. During daytime, hydroxyl radicals initiate monoterpene oxidation, mainly by hydroxyl addition to monoterpene double bonds. Naturally, related HOM formation mechanisms should be induced by that reaction route, too. However, for α-pinene, the most abundant atmospheric monoterpene, we find a previously unidentified competitive pathway under atmospherically relevant conditions: HOM formation is predominately induced via hydrogen abstraction by hydroxyl radicals, a generally minor reaction pathway. We show by observations and theoretical calculations that hydrogen abstraction followed by formation and rearrangement of alkoxy radicals is a prerequisite for fast daytime HOM formation. Our analysis provides an accurate mechanism and yield, demonstrating that minor reaction pathways can become major, here for SOA formation and growth and related impacts on air quality and climate.

APR-246 triggers ferritinophagy and ferroptosis of diffuse large B-cell lymphoma cells with distinct TP53 mutations.

TP53 mutations correlate with inferior survival in many cancers. APR-246 is a compound to shift mutant p53 and exhibits anti-cancer effects. Among its effects, APR-246 facilitates the binding of restored p53 mutants to target genes and their transcription. A set of 2464 DLBCL cases from multiple cohorts including our center, was integrated to identify the type and localization of TP53 mutations and clinical impacts. APR-246 was applied in TP53-mutated DLBCL cells and xenograft mouse models to explore the anti-tumor effect. TP53 mutations frequency was 16% and TP53 mutations correlated with poor overall survival (OS) and progression-free survival (PFS) in all cases, especially in germinal center B-cell-like (GCB) and unclassified (UNC) subtypes. Notably, TP53 single mutations in the DNA binding domain (DBD) led to poor OS and PFS. Specifically, mutations in exon 7 correlated with poorer OS, while mutations in exons 5 and 6 associated with inferior PFS. APR-246 induces p53-dependent ferritinophagy of DLBCL cells with TP53 missense mutation on exon 7 and ferroptosis of DLBCL cells harboring wild-type TP53 and other TP53 mutations. TP53 mutations on exons 5, 6 and 7 are predictors of progression and survival. Targeting mutant p53 by APR-246 is a promising therapeutic approach for DLBCL patients.

The integrated landscape of eRNA in gastric cancer reveals distinct immune subtypes with prognostic and therapeutic relevance.

The comprehensive regulation effect of eRNA on tumor immune cell infiltration and the outcome remains obscure. We comprehensively identify the eRNA-mediated immune infiltration patterns of gastric cancer (GC) samples. We creatively proposed a random forest machine-learning (ML) algorithm to map eRNA to mRNA expression patterns. The eRNA score was constructed using principal component analysis algorithms and validated in an independent cohort. Three subtypes with distinct eRNA expression patterns were determined in GC. There were significant differences between the three subtypes in the overall survival rate, immune cell infiltration characteristics, and immunotherapy response indicators. The patients in the high eRNA score group have a higher overall survival rate and might benefit from immunotherapy. This work revealed that eRNA regulation might be a new prognostic index and might offer a potential biomarker in the response of immunotherapy. Evaluating the eRNA regulation manner of GC will contribute to guiding more effective immunotherapy strategies.

A single-cell analysis reveals tumor heterogeneity and immune environment of acral melanoma.

Acral melanoma is a dismal subtype of melanoma occurring in glabrous acral skin, and has a higher incidence in East Asians. We perform single-cell RNA sequencing for 63,394 cells obtained from 5 acral and 3 cutaneous melanoma samples to investigate tumor heterogeneity and immune environment. We define 5 orthogonal functional cell clusters that are involved in TGF-beta signaling, Type I interferon, Wnt signaling, Cell cycle, and Cholesterol efflux signaling. Signatures of enriched TGF-beta, Type I interferon, and cholesterol efflux signaling are significantly associated with good prognosis of melanoma. Compared with cutaneous melanoma, acral melanoma samples have significantly severe immunosuppressive state including depletion of cytotoxic CD8+ T cells, enrichment of Treg cells, and exhausted CD8+ T cells. PD1 and TIM-3 have higher expression in the exhaustive CD8+ T cells of acral melanoma. Key findings are verified in two independent validation sets. This study contributes to our better understanding of acral melanoma.

Deep learning to diagnose Hashimoto's thyroiditis from sonographic images.

Hashimoto's thyroiditis (HT) is the main cause of hypothyroidism. We develop a deep learning model called HTNet for diagnosis of HT by training on 106,513 thyroid ultrasound images from 17,934 patients and test its performance on 5051 patients from 2 datasets of static images and 1 dataset of video data. HTNet achieves an area under the receiver operating curve (AUC) of 0.905 (95% CI: 0.894 to 0.915), 0.888 (0.836-0.939) and 0.895 (0.862-0.927). HTNet exceeds radiologists' performance on accuracy (83.2% versus 79.8%; binomial test, p < 0.001) and sensitivity (82.6% versus 68.1%; p < 0.001). By integrating serologic markers with imaging data, the performance of HTNet was significantly and marginally improved on the video (AUC, 0.949 versus 0.888; DeLong's test, p = 0.004) and static-image (AUC, 0.914 versus 0.901; p = 0.08) testing sets, respectively. HTNet may be helpful as a tool for the management of HT.

Prediction of Two Molecular Subtypes of Gastric Cancer Based on Immune Signature.

Gastric cancer is the fifth most common type of human cancer and the third leading cause of cancer-related death. The purpose of this study is to investigate the immune infiltration signatures of gastric cancer and their relation to prognosis. We identified two distinct subtypes of gastric cancer (C1/C2) characterized by different immune infiltration signatures. C1 is featured by immune resting, epithelial-mesenchymal transition, and angiogenesis pathways, while C2 is featured by enrichment of the MYC target, oxidative phosphorylation, and E2F target pathways. The C2 subtype has a better prognosis than the C1 subtype (HR = 0.61, 95% CI: 0.44-0.85; log-rank test, p = 0.0029). The association of C1/C2 with prognosis remained statistically significant (HR = 0.62, 95% CI: 0.44-0.87; p = 0.006) after controlling for age, gender, and stage. The prognosis prediction of C1/C2 was verified in four independent cohorts (including an internal cohort). In summary, our study is helpful for better understanding of the association between immune infiltration and the prognosis of gastric cancer.

Transcriptomic Analysis Identified Two Subtypes of Brain Tumor Characterized by Distinct Immune Infiltration and Prognosis.

BACKGROUND: Brain tumor ranks as the most devastating cancer type. The complex tumor immune microenvironment prevents brain tumor from receiving therapeutic benefits. The purpose of this study was to stratify brain tumors based on their distinct immune infiltration signatures to facilitate better clinical decision making and prognosis prediction. METHODS: We developed a deep learning model to characterize immune infiltration from transcriptome. The developed model was applied to distill expression signatures of transcriptome of brain tumor samples. We performed molecular subtyping with the extracted expression signatures to unveil brain tumor subtypes. Computational methods, including gene set enrichment analysis, Kaplan-Meier survival and multivariate Cox regression analyses, were employed. RESULTS: We identified two distinctive subtypes (i.e. C1/2) of brain tumor featured by distinct immune infiltration signatures. The C1 subtype is characterized by protective immune infiltration signatures, including high infiltration of CD8+ T cells and activation of CX3CL1. The C2 subtype has an extensive infiltration of tumor-associated macrophages and microglia, and was enriched with immune suppressive, wound-healing, and angiogenic signatures. The C1 subtype had significantly better prognosis as compared with C2 (Log-rank test, HR: 2.5, 95% CI: 2.2 - 2.7; P = 8.2e-78). This difference remained statistically significant (multivariate Cox model, HR: 2.2, 95% CI: 1.7 - 2.9; P = 3.7e-10) by taking into account age, gender, recurrent/secondary status at sampling time, tumor grade, histology, radio-chemotherapy, IDH mutation, MGMT methylation, and co-deletion of 1p and 19q. This finding was validated in six datasets. The C2 subtype of glioblastoma patients with IDH mutation has poor survival analogous to those without IDH mutation (Log-rank test, adjusted P = 0.8), while C1 has favorable prognosis as compared with glioblastoma of C2 subtype with IDH mutation (Log-rank test, adjusted P = 1.2e-3) or without IDH mutation (Log-rank test, adjusted P = 1.3e-6). CONCLUSIONS: We identified two distinctive subtypes of brain tumor with different immune infiltration signatures, which might be helpful as an independent prognosticator for brain tumor.