Search PubMed⌕ Search

SEARCH · Search PubMed

Results for “Cell annotation”

Search indexed PubMed citations on genomics, clinical trials, systematic reviews and public health. Explore titles, authors and supplied subject terms, then open the PubMed record.

Quote a phrase for an exact phrase match. Source license links do not imply unrestricted reuse.

At least 469 records · Page 26Linked to original sources

Transcriptional analysis of latent and inducible Kaposi's sarcoma-associated herpesvirus transcripts in the K4 to K7 region.

Kaposi's sarcoma-associated herpesvirus (KSHV) is a gamma-2 herpesvirus with a genome containing a long unique coding region (LUR) flanked by GC-rich terminal repeat sequences. The LUR encodes approximately 90 annotated open reading frames (ORFs) with complex patterns of gene expression during viral latency, reactivation, and de novo infection. To identify unannotated KSHV genes, we examined the region between 21,500 and 30,000 bp of the KSHV LUR, representing approximately 8.5 kb of sequence. This region encodes seven known single-exon ORFs (K4, K4.1, K4.2, K5, K6, K7, and PAN), but previous computer analyses have failed to identify additional likely genes in the remaining 5.2 kb. We identified four novel transcripts using Northern blotting, phage library screening, and 5' rapid amplification of cDNA ends analysis in the region between ORFs K4.2 and K7. In vitro analysis of KSHV-infected primary effusion lymphoma cell lines in the presence of 12-O-tetradecanoylphorbol-13-acetate and phosphonoformic acid suggests that one latent transcript is coterminal with the previously annotated K3 gene encoding an ubiquitin-ligase known to downregulate major histocompatibility complex class I expression. This alternatively spliced transcript may contribute to KSHV adaptive immune evasion during latent infection. Other transcripts are inducible, including a 6.1-kb transcript that is the largest transcript found in the KSHV genome to date.

Cell Line↗

Immunogenetics Sequence Annotation: the Strategy of IMGT based on IMGT-ONTOLOGY.

IMGT, the international ImMunoGeneTics information system((R))(http://imgt.cines.fr) created in 1989, by the Laboratoire d'ImmunoGénétique Moléculaire (LIGM), Université Montpellier II and CNRS, Montpellier, France, is a high quality integrated information system, secialized in immunoglobulins (IG), T cell receptors (TR), major histocompatibility complex of human and other vertebrates and related proteins of the immune system that belong to the IgSF and Mhc superfamilies. IMGT/LIGM-DB, the first and the largest IMGT database, manages more than 92,000 IG and TR nucleotide sequences from human and 150 other vertebrate species in May 2005. IMGT/LIGM-DB provides expertly annotated sequences and standardized knowledge based on IMGT-ONTOLOGY, the first ontology for immunogenetics and immunoinformatics. The strategy developed by IMGT, for the IG and TR nucleotide sequence annotation, involves two different approaches that depend on the nature of the sequences, genomic DNA (gDNA) or complementary DNA (cDNA).

Animals↗

Proteome of Methanosarcina acetivorans Part II: comparison of protein levels in acetate- and methanol-grown cells.

Methanosarcina acetivorans is an archaeon isolated from marine sediments which utilizes a diversity of substrates for growth and methanogenesis. Part I of a two-part investigation has profiled proteins of this microorganism cultured with both methanol and acetate as growth substrates, utilizing two-dimensional gel electrophoresis and MALDI-TOF-TOF mass spectrometry. In this report, Part II, the analyses were extended to identify 34 proteins found to be present in different amounts between methanol- and acetate-grown M. acetivorans. Among these proteins are enzymes which function in pathways for methanogenesis from either acetate or methanol. Several of the 34 proteins were determined to have redundant functions based on annotations of the genomic sequence. Enzymes which function in ATP synthesis and steps common to both methanogenic pathways were elevated in acetate- versus methanol-grown cells, whereas enzymes that have a more general function in protein synthesis were in greater amounts in methanol- compared to acetate-grown cells. Several group I chaperonins were present in greater amounts in methanol- versus acetate-grown cells, whereas lower amounts of several stress related proteins were found in methanol- versus acetate-grown cells. The potential physiological basis for these novel patterns of protein synthesis are discussed.

Acetates↗

Comparative analysis of convolutional neural network models for the histopathological differentiation of acinic cell carcinoma and secretory carcinoma.

OBJECTIVE: Although artificial intelligence tools show promise for enhancing the diagnosis of head and neck lesions, few studies have tested these resources for the microscopic diagnosis of salivary gland tumors. Specifically, the microscopic differentiation between acinic cell carcinoma and secretory carcinoma has never been addressed in this context. Therefore, this exploratory study aimed to comparatively evaluate the feasibility of applying convolutional neural networks for the microscopic differentiation between acinic cell carcinomas and secretory carcinomas. METHODS: A cross-sectional study using whole-slide images from 46 patients with acinic cell carcinoma (n = 26) or secretory carcinoma (n = 20) was conducted. Eight CNNs (ResNet-50, InceptionV3, VGG16, Xception, MobileNet, DenseNet121, EfficientNetB0, and EfficientNetV2B0) were trained and evaluated for accuracy, sensitivity, specificity, F1-score, and AUC. Performance was measured in training, validation, and test subsets. Accuracy and loss curves were also presented. RESULTS: InceptionV3 demonstrated the best overall performance, with the lowest loss (1.39), highest accuracy (0.81), sensitivity (0.90), and F1-score (0.81). VGG16 achieved the highest AUC (0.86) and precision (0.77). DenseNet121 showed the lowest performance in terms of accuracy (0.65) and F1-Score (0.52), but the highest specificity (0.85). CONCLUSION: This proof-of-concept study suggests that convolutional neural networks may be feasible tools to support the microscopic differentiation between acinic cell carcinoma and secretory carcinoma. The performance of these models critically depends on the size of the dataset and the quality of annotations. The findings should be interpreted cautiously given the limited dataset and potential sources of bias. Further validation with larger, multicenter datasets is needed before any clinical application can be considered.

Humans↗

Zinc finger proteins and other transcription regulators as response proteins in benzo[a]pyrene exposed cells.

Proteomic analysis, which combines two-dimensional electrophoresis (2-DE) and mass spectrometry (MS), is an important approach to screen proteins responsive to specific stimuli. Benzo[a]pyrene (B[a]P), a prototype of polycyclic hydrocarbons (PAHs), is a potent procarcinogen generated from the combustion of fossil fuel and cigarette smoke. To further probe the molecular mechanism of mutagenesis and carcinogenesis, and to find potential molecular markers involved in cellular responses to B[a]P exposure, we performed proteomic analysis of whole cellular proteins in human amnion epithelial cells after B[a]P-treatment. Image visualization and statistical analysis indicated that more than 40 proteins showed significant changes following B[a]P-treatment (P < 0.05). Among them, 20 proteins existed only in the control groups, while six were only present in B[a]P-treated cells. In addition, the expression of 10 proteins increased whereas 11 decreased after B[a]P-treatment. These proteins were subjected to in-gel tryptic digestion followed by matrix-assisted laser desorption/ionization-time of flight mass spectrometry (MALDI-TOF-MS) analysis. Using peptide mass fingerprinting (PMF) to search the nrNCBI database, we identified 22 proteins. Most of these proteins have unknown functions and have not been previously connected to a response to B[a]P exposure. To further annotate the characteristics of these proteins, GOblet analysis was carried out and results indicated that they were involved in multiple biological processes including regulation of transcription, cell proliferation, cell aging and other processes. However, expression changes were noted in a number of transcription regulators, including eight zinc finger proteins as well as SNF2L1 (SWI/SNF related, matrix associated, actin dependent regulator of chromatin, subfamily a, member 1), which is closely linked to the chromatin remodeling process. These data may provide new clues to further understand the implication of these proteins in cellular responses to carcinogen exposure as well as the molecular mechanisms of B[a]P-induced mutagenesis and carcinogenesis.

Amino Acid Sequence↗

Assembly, verification, and initial annotation of the NIA mouse 7.4K cDNA clone set.

A set of 7407 cDNA clones (NIA mouse 7.4K) was assembled from >20 cDNA libraries constructed mainly from early mouse embryos, including several stem cell libraries. The clone set was assembled from embryonic and newborn organ libraries consisting of ~120,000 cDNA clones, which were initially re-arrayed into a set of ~11,000 unique cDNA clones. A set of tubes was constructed from the racks in this set to prevent contamination and potential mishandling errors in all further re-arrays. Sequences from this set (11K) were analyzed further for quality and clone identity, and high-quality clones with verified identity were re-arrayed into the final set (7.4K). The set is freely available, and a corresponding database was built to provide comprehensive annotation for those clones with known identity or homology, and has been made available through an extensive Web site that includes many link-outs to external databases and analysis servers.

Animals↗

A high-resolution map of active promoters in the human genome.

In eukaryotic cells, transcription of every protein-coding gene begins with the assembly of an RNA polymerase II preinitiation complex (PIC) on the promoter. The promoters, in conjunction with enhancers, silencers and insulators, define the combinatorial codes that specify gene expression patterns. Our ability to analyse the control logic encoded in the human genome is currently limited by a lack of accurate information regarding the promoters for most genes. Here we describe a genome-wide map of active promoters in human fibroblast cells, determined by experimentally locating the sites of PIC binding throughout the human genome. This map defines 10,567 active promoters corresponding to 6,763 known genes and at least 1,196 un-annotated transcriptional units. Features of the map suggest extensive use of multiple promoters by the human genes and widespread clustering of active promoters in the genome. In addition, examination of the genome-wide expression profile reveals four general classes of promoters that define the transcriptome of the cell. These results provide a global view of the functional relationships among transcriptional machinery, chromatin structure and gene expression in human cells.

Chromatin↗

Identification of new human cadherin genes using a combination of protein motif search and gene finding methods.

We have combined protein motif search and gene finding methods to identify genes encoding proteins containing specific domains. Particularly, we have focused on finding new human genes of the cadherin superfamily proteins, which represent a major group of cell-cell adhesion receptors contributing to embryonic neuronal morphogenesis. Models for three cadherin protein motifs were generated from over 100 already annotated cadherin domains and used to search the complete translated human genome. The genomic sequence regions containing motif "hits" were analyzed by eukaryotic GeneMark.hmm to identify the exon-intron structure of new genes. Three new genes CDH-J, PCDH-J and FAT-J were found. The predicted proteins PCDH-J and FAT-J were classified into protocadherin and FAT-like subfamilies, respectively, based on the number and organization of cadherin domains and presence of subfamily-specific conserved amino acid residues. Expression of FAT-J was shown in almost all tested tissues. The exon-intron organization of CDH-J was experimentally verified by PCR with specifically designed primers and its tissue-specific expression was demonstrated. The described methodology can be applied to discover new genes encoding proteins from families with well-characterized structural and functional domains.

Amino Acid Motifs↗

High-Quality Genome Assembly, Metabolome, Pangenome, and Metabolic Models of Megasphaera hexanoica KCCM 43214T.

Megasphaera hexanoica KCCM 43214T, isolated from cow rumen, is capable of producing medium-chain carboxylic acids such as hexanoate and octanoate. In this study, we present a high-quality genome assembly, along with intracellular metabolomic profiling and pangenomic analysis. Illumina sequencing generated 2.3 Gbp from 15,293,634 reads with a GC content of 49.5%, while PacBio HiFi sequencing produced 331.5 Mbp across 45,266 reads, with an average read length of 7,323&#x2009;bp and a HiFi read N50 of 8,214&#x2009;bp. Hybrid assembly of short and long reads resulted in a single 2.88 Mbp contig, containing 2,835 protein-coding genes. Genome-scale metabolic models were constructed to evaluate its metabolic capabilities under specific growth conditions. Intracellular metabolomic analysis of cells grown in medium containing fructose and lactate revealed key metabolic activities associated with chain elongation. Pangenomic analysis across nine annotated genomes identified 6,721 orthologous genes using OrthoMCL, emphasizing the genetic and functional diversity within the Megasphaera genus. This dataset offers valuable insights into the metabolism and biotechnological potential of M. hexanoica KCCM 43214T.

Metabolome↗

Electron transport in the pathway of acetate conversion to methane in the marine archaeon Methanosarcina acetivorans.

A liquid chromatography-hybrid linear ion trap-Fourier transform ion cyclotron resonance mass spectrometry approach was used to determine the differential abundance of proteins in acetate-grown cells compared to that of proteins in methanol-grown cells of the marine isolate Methanosarcina acetivorans metabolically labeled with 14N versus 15N. The 246 differentially abundant proteins in M. acetivorans were compared with the previously reported 240 differentially expressed genes of the freshwater isolate Methanosarcina mazei determined by transcriptional profiling of acetate-grown cells compared to methanol-grown cells. Profound differences were revealed for proteins involved in electron transport and energy conservation. Compared to methanol-grown cells, acetate-grown M. acetivorans synthesized greater amounts of subunits encoded in an eight-gene transcriptional unit homologous to operons encoding the ion-translocating Rnf electron transport complex previously characterized from the Bacteria domain. Combined with sequence and physiological analyses, these results suggest that M. acetivorans replaces the H2-evolving Ech hydrogenase complex of freshwater Methanosarcina species with the Rnf complex, which generates a transmembrane ion gradient for ATP synthesis. Compared to methanol-grown cells, acetate-grown M. acetivorans synthesized a greater abundance of proteins encoded in a seven-gene transcriptional unit annotated for the Mrp complex previously reported to function as a sodium/proton antiporter in the Bacteria domain. The differences reported here between M. acetivorans and M. mazei can be attributed to an adaptation of M. acetivorans to the marine environment.

Acetates↗

Characterization of the putative operon containing arylamine N-acetyltransferase (nat) in Mycobacterium bovis BCG.

Mycobacterium bovis BCG and Mycobacterium tuberculosis possess a single arylamine N-acetyltransferase whose gene is predicted to occur within a six-gene operon. Deletion of the nat gene caused an extended lag phase in M. bovis BCG and a cell morphology associated with an altered pattern of cell wall mycolates. Analysis of cDNA from M. bovis BCG shows that during in vitro growth all the genes in the putative nat operon are expressed and the open reading frames are contiguous, supporting the existence of an operon. Two genes in the operon, Mb3599c and Mb3600c, are predicted to encode homologues of enzymes annotated as a 2,3-dihydroxybiphenyl 1,2-dioxygenase (bphC5) and a 2-hydroxy-6-oxo-6-phenylhexa-2,4-dienoate hydrolase (bphD2), respectively, in Rhodococcus RHA1. As predicted, M. bovis BCG cell lysates metabolized the BphC substrate 2,3-dihydroxybiphenyl (2,3-DHB) to 2-hydroxy-6-oxo-6-phenylhexa-2,4-dienoic acid (HOPDA), a BphD substrate, which was subsequently hydrolysed. Immunoprecipitation of the BphD homologue from these lysates led to an accumulation of HOPDA. M. bovis BCG growth on both solid and liquid media was inhibited with either 2,3-DHB or an inhibitor of BphC, 3-chlorocatechol (3-CC). In addition, incubation with 2,3-DHB affects the lipid composition of the cell wall resulting in a diminished level of mycolates and an altered cell morphology similar to the Deltanat strain. We propose the enzymes encoded by the putative operon have a similar endogenous role to that of the NAT enzyme and are part of a pathway important for cell wall synthesis.

Arylamine N-Acetyltransferase↗

Receptor-defined targeting of a genomically unique melanoma-enriched noncanonical antigen.

Effective T cell-based immunotherapies require functional receptors that can be engineered and redeployed to recognize tumor-restricted antigens. Noncanonical peptides arising from transcription outside annotated protein-coding regions expand the antigenic landscape of cancer; however, systematic strategies to biologically prioritize and functionally validate such targets remain underdeveloped. Here, we integrated de novo transcript analysis, exon-resolved quantification, RNA in situ hybridization, and immunopeptidomics to identify melanoma-associated noncanonical transcripts and advance candidates through receptor-level validation. Among three recurrent melanoma-associated transcripts, EVA003 emerged as a lead target based on its distinct repeat-enriched genomic architecture, consistent tumor-enriched exon-level expression across independent datasets, and a genomically unique immunogenic core sequence. We demonstrate endogenous presentation of EVA003-derived peptides on HLA-A*03:01 and detect specific reactivity in patient-derived tumor-infiltrating lymphocytes. Single-cell transcriptomic profiling identified a dominant peptide-reactive clonotype, enabling isolation of a naturally occurring T cell receptor. Transfer of this receptor into healthy donor T cells conferred antigen-dependent activation and cytotoxicity against both peptide-pulsed targets and melanoma cells expressing EVA003 endogenously. Together, these findings establish a biologically informed strategy for prioritizing noncanonical tumor antigens and demonstrate that genomically unique, tumor-enriched noncanonical peptides can be presented to molecularly defined receptors capable of mediating cancer cell killing. These findings support the integration of prioritized noncanonical antigens into engineered T cell therapeutic strategies.

Humans↗

Three-way translocation involves MLL, MLLT3, and a novel cell cycle control gene, FLJ10374, in the pathogenesis of acute myeloid leukemia with t(9;11;19)(p22;q23;p13.3).

The MLL gene, at 11q23, undergoes chromosomal translocation with a large number of partner genes in both acute lymphoblastic and acute myeloid leukemia (AML). We report a novel t(9;11;19)(p22;q23;p13.3) disrupting MLL in an infant AML patient. The 5' end of MLL fused to chromosome 9 sequences on the der(11), whereas the 3' end was translocated to chromosome 19. We developed long-distance inverse-polymerase chain reaction assays to investigate the localization of the breakpoints on der(11) and der(19). We found that intron 5 of MLL was fused to intron 5 of MLLT3 at the der(11) genomic breakpoint, resulting in a novel in-frame MLL exon 5-MLLT3 exon 6 fusion transcript. On the der(19), a novel gene annotated as FLJ10374 was disrupted by the breakpoint. Using reverse transcription-polymerase chain reaction analysis, we showed that FLJ10374 is ubiquitously expressed in human cells. Transfection of the FLJ10374 protein in different cell lines revealed that it localized exclusively to the nucleus. In serum-starved NIH-3T3 cells, the expression of FLJ10374 decreased the rate of the G1-to-S transition of the cell cycle, whereas the suppression of FLJ10374 through short interfering RNA increased cell proliferation. These results indicate that FLJ10374 negatively regulates cell cycle progression and proliferation. Thus, a single chromosomal rearrangement resulting in formation of the MLL-MLLT3 fusion gene and haplo-insufficiency of FLJ10374 may have cooperated to promote leukemogenesis in AML with t(9;11;19).

Acute Disease↗

Genome-wide mapping of DNase hypersensitive sites using massively parallel signature sequencing (MPSS).

A major goal in genomics is to understand how genes are regulated in different tissues, stages of development, diseases, and species. Mapping DNase I hypersensitive (HS) sites within nuclear chromatin is a powerful and well-established method of identifying many different types of regulatory elements, but in the past it has been limited to analysis of single loci. We have recently described a protocol to generate a genome-wide library of DNase HS sites. Here, we report high-throughput analysis, using massively parallel signature sequencing (MPSS), of 230,000 tags from a DNase library generated from quiescent human CD4+ T cells. Of the tags that uniquely map to the genome, we identified 14,190 clusters of sequences that group within close proximity to each other. By using a real-time PCR strategy, we determined that the majority of these clusters represent valid DNase HS sites. Approximately 80% of these DNase HS sites uniquely map within one or more annotated regions of the genome believed to contain regulatory elements, including regions 2 kb upstream of genes, CpG islands, and highly conserved sequences. Most DNase HS sites identified in CD4+ T cells are also HS in CD8+ T cells, B cells, hepatocytes, human umbilical vein endothelial cells (HUVECs), and HeLa cells. However, approximately 10% of the DNase HS sites are lymphocyte specific, indicating that this procedure can identify gene regulatory elements that control cell type specificity. This strategy, which can be applied to any cell line or tissue, will enable a better understanding of how chromatin structure dictates cell function and fate.

Cell Differentiation↗

Analysis of bovine mammary gland EST and functional annotation of the Bos taurus gene index.

Functional genomic studies of the mammary gland require an appropriate collection of cDNA sequences to assess gene expression patterns from the different developmental and operational states of underlying cell types. To better capture the range of gene expression, a normalized cDNA library was constructed from pooled bovine mammary tissues, and 23,202 expressed sequence tags (EST) were produced and deposited into GenBank. Assembly of these EST with sequences in the Bos taurus Gene Index (BtGI) helped to form 5751 of the current 23,883 tentative consensus (TC) sequences. The majority (87%) of these 5751 assemblies contained only one to three mammary-derived EST. In contrast, 18% of the mammary EST assembled with TC sequences corresponding to 12 genes. These results suggest library normalization was only partially effective, because the reduction in EST for genes abundantly transcribed during lactation could be attributed to pooling. For better assessment of novel content in the mammary library and to add to existing annotation of all bovine sequence elements, gene ontology assignments, and comparative sequence analyses against human genome sequence, human and rodent gene indices, and an index of orthologous alignments of genes across eukaryotes (TOGA) were performed, and results were added to existing BtGI annotation. Over 35,000 of the bovine elements significantly matched human genome sequence, and the positions of some alignments (3%) were unique relative to those using human expressed sequences. Because 3445 TC sequences had no significant match with any data set, mammary-derived cDNA clones representing 23 of these elements were analyzed further for expression and novelty. Only one clone met criteria suggesting the corresponding gene was a divergent ortholog or expressed sequence unique to cattle. These results demonstrate that bovine sequence expression data serve as a resource for characterizing mammalian transcriptomes and identifying those genes potentially unique to ruminants.

Animals↗

Protein complexes and functional modules in molecular networks.

Proteins, nucleic acids, and small molecules form a dense network of molecular interactions in a cell. Molecules are nodes of this network, and the interactions between them are edges. The architecture of molecular networks can reveal important principles of cellular organization and function, similarly to the way that protein structure tells us about the function and organization of a protein. Computational analysis of molecular networks has been primarily concerned with node degree [Wagner, A. & Fell, D. A. (2001) Proc. R. Soc. London Ser. B 268, 1803-1810; Jeong, H., Tombor, B., Albert, R., Oltvai, Z. N. & Barabasi, A. L. (2000) Nature 407, 651-654] or degree correlation [Maslov, S. & Sneppen, K. (2002) Science 296, 910-913], and hence focused on single/two-body properties of these networks. Here, by analyzing the multibody structure of the network of protein-protein interactions, we discovered molecular modules that are densely connected within themselves but sparsely connected with the rest of the network. Comparison with experimental data and functional annotation of genes showed two types of modules: (i) protein complexes (splicing machinery, transcription factors, etc.) and (ii) dynamic functional units (signaling cascades, cell-cycle regulation, etc.). Discovered modules are highly statistically significant, as is evident from comparison with random graphs, and are robust to noise in the data. Our results provide strong support for the network modularity principle introduced by Hartwell et al. [Hartwell, L. H., Hopfield, J. J., Leibler, S. & Murray, A. W. (1999) Nature 402, C47-C52], suggesting that found modules constitute the "building blocks" of molecular networks.

Biophysical Phenomena↗

Exploration of genes showing intramuscular fat deposition-associated expression changes in musculus longissimus muscle.

Marbling, as defined by the amount of intramuscular fat, is an economically important trait in beef cattle. Intramuscular fat deposition is postulated to arise mainly from a series of adipogenic events in intramuscular adipocyte-lineage cells and in the physiological or anatomical milieux surrounding them. This study was designed to investigate gene-expression patterns associated with fat deposition in musculus longissimus muscle, including adipocyte-lineage cells and part of the milieux. Differential-display PCR (ddPCR) was used to examine expression differences between low-marbled and high-marbled steer groups at 8, 10, 12 and 14 months of age, encompassing the time that marbling starts to appear. Seventy-four of 2114 total bands on ddPCR gel-bands were significant (P < 0.05) for the group effect, the interaction effect between group and age, or both the group and the interaction effects. Sequence analysis of 72 of these bands revealed 77 genes, including 35 annotated genes and 42 novel sequences. Among the 35 annotated genes, 6 (BTG2, PDHB, SORBS1, TRDN, TTN and MGP) have been related to changes in intramuscular fat deposition, possibly by exerting effects on adipocyte-lineage cells or on the milieux surrounding them.

Adipose Tissue↗

Hierarchical modeling of tumor subtypes in cell lines using large-scale genomic datasets.

Cancer cell lines (CLs) are widely used to study tumor biology and drug response, yet their translational relevance is often limited by inaccurate subtype annotations. Existing CL-tumor matching approaches are frequently constrained by flat classification schemes, weak subtype definitions, and the exclusion of normal tissue references, leading to potential confounding of tumor-specific and tissue-of-origin signals. To address these limitations, a hierarchical classification (HC) framework is presented in which CLs are aligned with patient tumors across biological resolutions, from organ to molecular subtype. Gene expression profiles from 802 CLs, 5,612 tumors from The Cancer Genome Atlas (TCGA) , and 8,939 non-cancerous tissues were integrated to separate oncogenic signals from tissue-specific signals. Node-specific features were selected using maximum relevance minimum redundancy, and balanced accuracies of 89% in cross-validation and 75%, and 80% on external datasets were achieved. Through the framework, 43 CLs were reassigned, and clinically relevant underrepresented subtypes were identified.

cancer cell lines↗