High-content microscopy and machine learning characterize a cell morphology signature ofNF1genotype in Schwann cells DOI Creative Commons
Jenna Tomkinson,

Cameron Mattson,

Michelle Mattson‐Hoss

и другие.

bioRxiv (Cold Spring Harbor Laboratory), Год журнала: 2024, Номер unknown

Опубликована: Сен. 16, 2024

Abstract Neurofibromatosis type 1 (NF1) is a multi-system, autosomal dominant genetic disorder driven by the systemic loss of NF1 protein neurofibromin. Loss neurofibromin in Schwann cells particularly detrimental, as acquisition ‘second-hit’ (e.g., complete NF1) can lead to development plexiform neurofibroma tumors. Plexiform neurofibromas are painful, disfiguring tumors with an approximately 5 chance sarcoma transition. Selumetinib currently only medicine approved U.S. Food and Drug Administration (FDA) for treatment subset patients. This motivates need develop new therapies, either derived treat haploinsufficiency or function. To identify we understand impact has on cells. Here, aimed characterize differences high-content microscopy imaging neurofibromin-deficient We applied fluorescence assay (called Cell Painting) two isogenic cell lines, one wildtype genotype ( +/+ ) null -/- ). modified canonical Painting mark four organelles/subcellular compartments: nuclei, endoplasmic reticulum, mitochondria, F-actin. utilized CellProfiler pipelines perform quality control, illumination correction, segmentation, morphology feature extraction. segmented 22,585 cells, 907 significant features representing various organelle shapes intensity patterns, trained logistic regression machine learning model predict single The had high performance, training testing data yielding balanced accuracy 0.85 0.80, respectively. All our processing analyses freely available GitHub. look improve upon this preliminary future applying it large-scale drug screens deficient candidate drugs that return patient phenocopy healthier phenotype.

Язык: Английский

A versatile information retrieval framework for evaluating profile strength and similarity DOI Creative Commons
Alexandr A. Kalinin, John Arévalo, Loan Vulliard

и другие.

bioRxiv (Cold Spring Harbor Laboratory), Год журнала: 2024, Номер unknown

Опубликована: Апрель 2, 2024

Abstract In profiling assays, thousands of biological properties are measured in a single test, yielding discoveries by capturing the state cell population, often at single-cell level. However, for datasets, it has been challenging to evaluate phenotypic activity sample and consistency among samples, due profiles’ high dimensionality, heterogeneous nature, non-linear properties. Existing methods leave researchers uncertain where draw boundaries between meaningful response technical noise. Here, we developed statistical framework that uses well-established mean average precision (mAP) as single, data-driven metric bridge this gap. We validated mAP against established metrics through simulations real-world data applications, revealing its ability capture subtle differences state. Specifically, used assess both given perturbation (or sample) well within groups perturbations samples) across diverse high-dimensional datasets. evaluated on different profile types (image, protein, mRNA profiles), (CRISPR gene editing, overexpression, small molecules), resolutions (single-cell bulk). Our open-source software allows be applied identify interesting phenomena promising therapeutics from large-scale data.

Язык: Английский

Процитировано

6
Making the most of bioimaging data through interdisciplinary interactions DOI Creative Commons
Virginie Uhlmann, Matthew Hartley, Josh Moore

и другие.

Journal of Cell Science, Год журнала: 2024, Номер 137(20)

Опубликована: Окт. 15, 2024

The increasing technical complexity of all aspects involving bioimages, ranging from their acquisition to analysis, has led a diversification in the expertise scientists engaged at different stages discovery process. Although this diversity profiles comes with major challenge establishing fruitful interdisciplinary collaboration, such collaboration also offers superb opportunity for scientific discovery. In Perspective, we review actors within bioimaging research universe and identify primary obstacles that hinder interactions. We advocate data sharing, which lies heart innovation, is finally reach after decades being viewed as next impossible bioimaging. Building on recent community efforts, propose actions consolidate development truly culture based open exchange highlight promising outlook an example multidisciplinary endeavour.

Язык: Английский

Процитировано

1
High-content microscopy and machine learning characterize a cell morphology signature ofNF1genotype in Schwann cells DOI Creative Commons
Jenna Tomkinson,

Cameron Mattson,

Michelle Mattson‐Hoss

и другие.

bioRxiv (Cold Spring Harbor Laboratory), Год журнала: 2024, Номер unknown

Опубликована: Сен. 16, 2024

Abstract Neurofibromatosis type 1 (NF1) is a multi-system, autosomal dominant genetic disorder driven by the systemic loss of NF1 protein neurofibromin. Loss neurofibromin in Schwann cells particularly detrimental, as acquisition ‘second-hit’ (e.g., complete NF1) can lead to development plexiform neurofibroma tumors. Plexiform neurofibromas are painful, disfiguring tumors with an approximately 5 chance sarcoma transition. Selumetinib currently only medicine approved U.S. Food and Drug Administration (FDA) for treatment subset patients. This motivates need develop new therapies, either derived treat haploinsufficiency or function. To identify we understand impact has on cells. Here, aimed characterize differences high-content microscopy imaging neurofibromin-deficient We applied fluorescence assay (called Cell Painting) two isogenic cell lines, one wildtype genotype ( +/+ ) null -/- ). modified canonical Painting mark four organelles/subcellular compartments: nuclei, endoplasmic reticulum, mitochondria, F-actin. utilized CellProfiler pipelines perform quality control, illumination correction, segmentation, morphology feature extraction. segmented 22,585 cells, 907 significant features representing various organelle shapes intensity patterns, trained logistic regression machine learning model predict single The had high performance, training testing data yielding balanced accuracy 0.85 0.80, respectively. All our processing analyses freely available GitHub. look improve upon this preliminary future applying it large-scale drug screens deficient candidate drugs that return patient phenocopy healthier phenotype.

Язык: Английский

Процитировано

0