Deep learning of representations for transcriptomics-based phenotype prediction

Am, Smith; Walsh, Martin A.; Long, John M.; Cb, Davis; Henstock, Peter V.; Mr, Hodge; Maciejewski, Mateusz; Xj, Mu; S, Ra; Zhao, Shanrong; Ziemek, Daniel; Ck, Fisher

doi:10.1101/574723

Cited by 6 publications

(7 citation statements)

References 48 publications

(55 reference statements)

Supporting

Mentioning

Contrasting

Order By: Relevance

“…When we used tissue studies of the human pancreas [11][12][13][14] and the mouse brain [15][16][17][18] , all machine-learning models and cellFishing.jl performed equally well on the prediction task. Smith et al 25 reported a similar result in bulk RNA-seq data. In contrast, the single-cell specific classifiers scMatch and Garnett showed lower accuracy on the same datasets.…”

Section: Most Machine Learning Methods Are Well-suited For Cell Type Annotation Given Optimal Hyper-parameter Valuesmentioning

confidence: 55%

Deep learning does not outperform classical machine learning for cell-type annotation

Büttner

Andriamanga

et al. 2019

Preprint

View full text Add to dashboard Cite

Deep learning has revolutionized image analysis and natural language processing with remarkable accuracies in prediction tasks, such as image labeling or word identification. The origin of this revolution was arguably the deep learning approach by the Hinton lab in 2012, which halved the error rate of existing classifiers in the then 2year-old ImageNet database 1 . In hindsight, the combination of algorithmic and hardware advances with the appearance of large and well-labeled datasets has led up to this seminal contribution. The emergence of large amounts of data from single-cell RNA-seq and the recent global effort to chart all cell types in the Human Cell Atlas has attracted an interest in deep-learning applications. However, all current approaches are unsupervised, i.e., learning of latent spaces without using any cell labels, even though supervised learning approaches are often more powerful in feature learning and the most popular approach in the current AI revolution by far. Here, we ask why this is the case. In particular we ask whether supervised deep learning can be used for cell annotation, i.e. to predict cell-type labels from single-cell gene expression profiles. After evaluating 6 classification methods across 14 datasets, we notably find that deep learning does not outperform classical machine-learning methods in the task. Thus, cell-type prediction based on gene-signature derived celltype labels is potentially too simplistic a task for complex non-linear methods, which demands better labels of functional single-cell readouts. We, therefore, are still waiting for the "ImageNet moment" in single-cell genomics.

show abstract

Section: Most Machine Learning Methods Are Well-suited For Cell Type Annotation Given Optimal Hyper-parameter Valuesmentioning

confidence: 55%

Deep learning does not outperform classical machine learning for cell-type annotation

Büttner

Andriamanga

et al. 2019

Preprint

View full text Add to dashboard Cite

show abstract

“…Determination of the prediction that can confidently give rise to the gene expression in the body due to cancer can be done by deep learning process. As illustrated by Smith et al [12] that there are various traditional as well as scientific processes that can analyse the expression mechanism in the patients suffering from cancer. It has been analyzed that machine learning methods often provide a clear cut idea about the physiological and morphological changes that are occurring in the cells due to cancer.…”

Section: Critical Analysis Of the Deep Representation Of Learning On Phenotype Prediction From Gene Expressionmentioning

confidence: 99%

“…It has been analyzed that machine learning methods often provide a clear cut idea about the physiological and morphological changes that are occurring in the cells due to cancer. Smith et al [12] have also discussed the fact that genes are rarely active and act in isolation. It might be because along with the isolation procedure the physiological changes also happen in the genes in terms of its regulation.…”

Section: Critical Analysis Of the Deep Representation Of Learning On Phenotype Prediction From Gene Expressionmentioning

confidence: 99%

Deep Learning-based Models of Molecular Phenotypes for Predicting the Overall Survival in Cancer

Cazzato

Oak²,

Khan

et al. 2021

JPRI

View full text Add to dashboard Cite

Aims: The aim of the study is to justify the need of deep learning predictive model in obtaining molecular phenotypes of overall cancer survival. Study Design: The study is based on the secondary qualitative data analysis through usage of systematic review. Methodology: A qualitative study has been conducted to analyse the necessity of deep learning. It also includes the need for deep learning models to obtain the imaging of the cancer cells. In the study, a detailed discussion on deep learning has been made. The analysis of the primary sources has been obtained by evaluating the quality of the resources in the study. The study also comprises of a thematic analysis that enlightens the benefits of deep learning. The study is based on the analysis of 14 primary research-based articles out of 112 quantitative articles and structuring of a systematic review from the collected data. Results: The morphological and physiological changes that occur in the cancerous cells have been clearly evaluated in the research. The result signifies the prediction can be made by implementing deep learning in terms of cancer survival. Advancements in terms of technology in the medical field can thus be improved with the help of the deep learning process. It states the advancements of the deep learning models that are helpful in predicting the model of cancer to determine survival rate. Conclusion: Deep learning is a process that is considered to be a subset of artificial intelligence. Deep learning programmes are meant to be performed for complex learning models. Although there is difference in the concept of deep learning and image processing still artificial intelligence brings both together so as to ensure better performance in image processing. The need for deep learning models has become invasive, and it helps to build a strong ground for cancer survival.

show abstract

“…Deep learning may outperform other methods in big data classification with respect to accuracy of prediction [27], but not necessarily in other specific tasks, such as image-based cell-type annotation, for example [28]. It is not always clear which processing step in a deep learning approach would account for better results obtained [29].…”

Section: Big Data Analytics: From Machine Learning To Artificial Intementioning

confidence: 99%

Occam’s Razor for Big Data? On Detecting Quality in Large Unstructured Datasets

Dresp-Langley

Ekseth²,

Fesl

et al. 2019

Applied Sciences

View full text Add to dashboard Cite

Detecting quality in large unstructured datasets requires capacities far beyond the limits of human perception and communicability and, as a result, there is an emerging trend towards increasingly complex analytic solutions in data science to cope with this problem. This new trend towards analytic complexity represents a severe challenge for the principle of parsimony (Occam’s razor) in science. This review article combines insight from various domains such as physics, computational science, data engineering, and cognitive science to review the specific properties of big data. Problems for detecting data quality without losing the principle of parsimony are then highlighted on the basis of specific examples. Computational building block approaches for data clustering can help to deal with large unstructured datasets in minimized computation time, and meaning can be extracted rapidly from large sets of unstructured image or video data parsimoniously through relatively simple unsupervised machine learning algorithms. Why we still massively lack in expertise for exploiting big data wisely to extract relevant information for specific tasks, recognize patterns and generate new information, or simply store and further process large amounts of sensor data is then reviewed, and examples illustrating why we need subjective views and pragmatic methods to analyze big data contents are brought forward. The review concludes on how cultural differences between East and West are likely to affect the course of big data analytics, and the development of increasingly autonomous artificial intelligence (AI) aimed at coping with the big data deluge in the near future.

show abstract

Deep learning of representations for transcriptomics-based phenotype prediction

Cited by 6 publications

References 48 publications

Deep learning does not outperform classical machine learning for cell-type annotation

Deep learning does not outperform classical machine learning for cell-type annotation

Deep Learning-based Models of Molecular Phenotypes for Predicting the Overall Survival in Cancer

Occam’s Razor for Big Data? On Detecting Quality in Large Unstructured Datasets

Contact Info

Product

Resources

About