OptCluster : an R package for determining the optimal clustering algorithm and optimal number of clusters.

Sekula, Michael

doi:10.18297/etd/2147

Cited by 12 publications

(18 citation statements)

References 43 publications

Supporting

Mentioning

Contrasting

Unclassified

Order By: Relevance

“…To reveal which preprocessing and clustering methods best recapitulate the predicted number of clusters based on known cell identity, we applied eight cluster estimation algorithms (optCluster package (49)) on the DE0.05 data set (centered and scaled by contig, Ward.D2 and a correlation dissimilarity matrix; Fig. 2D).…”

Section: Resultsmentioning

confidence: 99%

“…We used the best performing transformations from the clustering analysis, i.e. data centered and scaled by gene and a correlation dissimilarity matrix, and 8 cluster determination indices provided by the optCluster package (49). We allowed a minimum of 2 and a maximum of 32 clusters for this and later cluster determination analyses.…”

Section: Resultsmentioning

confidence: 99%

“…Next, we performed cluster estimation using the optClust() function of the optCluster package (49). The algorithms used on each dataset varied by whether the data were counts or continuous.…”

Section: Methodsmentioning

confidence: 99%

See 2 more Smart Citations

Molecular Profiling to Infer Neuronal Cell Identity: Lessons from small ganglia of the Crab Cancer borealis

Northcutt

Kick

Otopalik

et al. 2019

Preprint

View full text Add to dashboard Cite

54Understanding circuit organization depends on identification of cell types. Recent advances in 55 transcriptional profiling methods have enabled classification of cell types by their gene 56 expression. While exceptionally powerful and high throughput, the ground-truth validation of 57 these methods is difficult: if cell type is unknown, how does one assess whether a given 58 analysis accurately captures neuronal identity? To shed light on the capabilities and limitations 59 of solely using transcriptional profiling for cell type classification, we performed two forms of 60 transcriptional profiling -RNA-seq and quantitative RT-PCR, in single, unambiguously identified 61 neurons from two small crustacean networks: the stomatogastric and cardiac ganglia. We then 62 combined our knowledge of cell type with unbiased clustering analyses and supervised machine 63 learning to determine how accurately functionally-defined neuron types can be classified by 64 expression profile alone. Our results demonstrate that expression profile is able to capture 65 neuronal identity most accurately when combined with multimodal information that allows for 66 post-hoc grouping so analysis can proceed from a supervised perspective. Solely unsupervised 67 clustering can lead to misidentification and an inability to distinguish between two or more cell 68 types. Therefore, our study supports the general utility of cell identification by transcriptional 69 profiling, but adds a caution: it is difficult or impossible to know under what conditions 70 transcriptional profiling alone is capable of assigning cell identity. Only by combining multiple 71 modalities of information such as physiology, morphology or innervation target can neuronal 72 identity be unambiguously determined. 73 74 3 SIGNIFICANCE STATEMENT 75Single cell transcriptional profiling has become a widespread tool in cell identification, 76 particularly in the nervous system, based on the notion that genomic information determines cell 77 identity. However, many cell type classification studies are unconstrained by other cellular 78 attributes (e.g., morphology, physiology). Here, we systematically test how accurately 79 transcriptional profiling can assign cell identity to well-studied anatomically-and functionally-80 identified neurons in two small neuronal networks. While these neurons clearly possess distinct 81 patterns of gene expression across cell types, their expression profiles are not sufficient to 82 unambiguously confirm their identity. We suggest that true cell identity can only be determined 83 by combining gene expression data with other cellular attributes such as innervation pattern, 84 morphology, or physiology. 85 86 87 88 89Unambiguous classification of neuronal cell types is a long-standing goal in 90 neuroscience with the aim to understand the functional components of the nervous system that 91give rise to circuits and, ultimately, behavior (1-6). Beyond that, agreement upon neuronal cell 92 types provides the opportunity to greatly increase reproduci...

show abstract

Section: Resultsmentioning

confidence: 99%

Section: Resultsmentioning

confidence: 99%

See 1 more Smart Citation

Molecular Profiling to Infer Neuronal Cell Identity: Lessons from small ganglia of the Crab Cancer borealis

Northcutt

Kick

Otopalik

et al. 2019

Preprint

View full text Add to dashboard Cite

show abstract

“…Para uma apresentação e descrição completa das medidas de validação, assim como dos algoritmos de estimação, ver Brock et al (2008). 20 A ordem do número ótimo de k, selecionado pelo algoritmo de ranking de agregação desenvolvido por Sekula et al (2015) foi: k = 2, 4, 6, 3, 7, 8 5, 9 e 10. Testou-se k = 4 e k = 6, sendo a opção pelo maior número de clusters defi nida em razão do objetivo do estudo de obter maior detalhe quanto às diferentes tipologias.…”

Section: Do Cluster Ao Fuzzy Clusterunclassified

Desenvolvimento e potencial de clusters criativos para as cidades médias brasileiras

Melo

Paiva

2016

Nova econ.

View full text Add to dashboard Cite

Resumo: Este estudo investiga a presença de clusters criativos nos municípios médios brasileiros em 2010 e baseia-se na metodologia de identificação proposta por Machado, Simões e Diniz (2013). A análise das cidades médias evidencia a criação de tipologias e diferenças nos padrões das aglomerações criativas. Ao isolar os efeitos dos grandes centros e das cidades periféricas, contribui-se para melhorar o entendimento e conhecimento da economia criativa no Brasil. O método multivariado de fuzzy cluster, que permite maior aproximação dos resultados com a realidade, foi aplicado a diferentes dados disponibilizados pelo IBGE, DATASUS e FINBRA. A partir de critérios específicos, foram selecionados seis clusters para a amostra. Analisou-se à disposição, localização geográfica e os resultados encontrados para cada cluster, sendo que quatro apresentaram um padrão mais aparente. A produção deste conhecimento é fonte importante para o exame adequado das políticas focadas na economia criativa e seus impactos.

show abstract

“…We apply several well-established machine learning techniques to address the technical challenges. First, to reduce dimensionality we utilize optCluster [21], an R package for determining the optimal clustering algorithm and optimal number of clusters. OptCluster identifies highly similar or repetitive expression patterns from genes, and clusters them into gene modules.…”

Section: Introductionmentioning

confidence: 99%

Leveraging TCGA gene expression data to build predictive models for cancer drug response

et al. 2020

View full text Add to dashboard Cite

Background Machine learning has been utilized to predict cancer drug response from multi-omics data generated from sensitivities of cancer cell lines to different therapeutic compounds. Here, we build machine learning models using gene expression data from patients’ primary tumor tissues to predict whether a patient will respond positively or negatively to two chemotherapeutics: 5-Fluorouracil and Gemcitabine. Results We focused on 5-Fluorouracil and Gemcitabine because based on our exclusion criteria, they provide the largest numbers of patients within TCGA. Normalized gene expression data were clustered and used as the input features for the study. We used matching clinical trial data to ascertain the response of these patients via multiple classification methods. Multiple clustering and classification methods were compared for prediction accuracy of drug response. Clara and random forest were found to be the best clustering and classification methods, respectively. The results show our models predict with up to 86% accuracy; despite the study’s limitation of sample size. We also found the genes most informative for predicting drug response were enriched in well-known cancer signaling pathways and highlighted their potential significance in chemotherapy prognosis. Conclusions Primary tumor gene expression is a good predictor of cancer drug response. Investment in larger datasets containing both patient gene expression and drug response is needed to support future work of machine learning models. Ultimately, such predictive models may aid oncologists with making critical treatment decisions.

show abstract

OptCluster : an R package for determining the optimal clustering algorithm and optimal number of clusters.

Abstract: Sekula, Michael N., "OptCluster : an R package for determining the optimal clustering algorithm and optimal number of clusters." (2015). Electronic Theses and Dissertations. Paper 2147.

Cited by 12 publications

References 43 publications

Molecular Profiling to Infer Neuronal Cell Identity: Lessons from small ganglia of the Crab Cancer borealis

Molecular Profiling to Infer Neuronal Cell Identity: Lessons from small ganglia of the Crab Cancer borealis

Desenvolvimento e potencial de clusters criativos para as cidades médias brasileiras

Leveraging TCGA gene expression data to build predictive models for cancer drug response

Contact Info

Product

Resources

About