| 1. |
- Fredin Haslum, Johan, et al.
(författare)
-
Cell Painting-based bioactivity prediction boosts high-throughput screening hit-rates and compound diversity
- 2024
-
Ingår i: Nature Communications. - : Springer Nature. - 2041-1723. ; 15:1
-
Tidskriftsartikel (refereegranskat)abstract
- Identifying active compounds for a target is a time- and resource-intensive task in early drug discovery. Accurate bioactivity prediction using morphological profiles could streamline the process, enabling smaller, more focused compound screens. We investigate the potential of deep learning on unrefined single-concentration activity readouts and Cell Painting data, to predict compound activity across 140 diverse assays. We observe an average ROC-AUC of 0.744 ± 0.108 with 62% of assays achieving ≥0.7, 30% ≥0.8, and 7% ≥0.9. In many cases, the high prediction performance can be achieved using only brightfield images instead of multichannel fluorescence images. A comprehensive analysis shows that Cell Painting-based bioactivity prediction is robust across assay types, technologies, and target classes, with cell-based assays and kinase targets being particularly well-suited for prediction. Experimental validation confirms the enrichment of active compounds. Our findings indicate that models trained on Cell Painting data, combined with a small set of single-concentration data points, can reliably predict the activity of a compound library across diverse targets and assays while maintaining high hit rates and scaffold diversity. This approach has the potential to reduce the size of screening campaigns, saving time and resources, and enabling primary screening with more complex assays.
|
|
| 2. |
- Fredin Haslum, Johan, et al.
(författare)
-
Bridging Generalization Gaps in High Content Imaging Through Online Self-Supervised Domain Adaptation
- 2024
-
Ingår i: Proceedings of the IEEE/CVF Winter Conference on Applications of Computer Vision (WACV), Waikoloa, HI, USA, 2024. ; , s. 7723-7732
-
Konferensbidrag (refereegranskat)abstract
- High Content Imaging (HCI) plays a vital role in modern drug discovery and development pipelines, facilitating various stages from hit identification to candidate drug characterization. Applying machine learning models to these datasets can prove challenging as they typically consist of multiple batches, affected by experimental variation, especially if different imaging equipment have been used. Moreover, as new data arrive, it is preferable that they are analyzed in an online fashion. To overcome this, we propose CODA, an online self-supervised domain adaptation approach. CODA divides the classifier’s role into a generic feature extractor and a task-specific model. We adapt the feature extractor’s weights to the new domain using cross-batch self-supervision while keeping the task-specific model unchanged. Our results demonstrate that this strategy significantly reduces the generalization gap, achieving up to a 300% improvement when applied to data from different labs utilizing different microscopes. CODA can be applied to new, unlabeled out-of-domain data sources of different sizes, from a single plate to multiple experimental batches.
|
|
| 3. |
- Fredin Haslum, Johan
(författare)
-
Machine Learning Methods for Image-based Phenotypic Profiling in Early Drug Discovery
- 2024
-
Doktorsavhandling (övrigt vetenskapligt/konstnärligt)abstract
- In the search for new therapeutic treatments, strategies to make the drug discovery process more efficient are crucial. Image-based phenotypic profiling, with its millions of pictures of fluorescent stained cells, is a rich and effective means to capture the morphological effects of potential treatments on living systems. Within this complex data await biological insights and new therapeutic opportunities – but computational tools are needed to unlock them.This thesis examines the role of machine learning in improving the utility and analysis of phenotypic screening data. It focuses on challenges specific to this domain, such as the lack of reliable labels that are essential for supervised learning, as well as confounding factors present in the data that are often unavoidable due to experimental variability. We explore transfer learning to boost model generalization and robustness, analyzing the impact of domain distance, initialization, dataset size, and architecture on the effectiveness of applying natural domain pre-trained weights to biomedical contexts. Building upon this, we delve into self-supervised pretraining for phenotypic image data, but find its direct application is inadequate in this context as it fails to differentiate between various biological effects. To overcome this, we develop new self-supervised learning strategies designed to enable the network to disregard confounding experimental noise, thus enhancing its ability to discern the impacts of various treatments. We further develop a technique that allows a model trained for phenotypic profiling to be adapted to new, unseen data without the need for any labels or supervised learning. Using this approach, a general phenotypic profiling model can be readily adapted to data from different sites without the need for any labels. Beyond our technical contributions, we also show that bioactive compounds identified using the approaches outlined in this thesis have been subsequently confirmed in biological assays through replication in an industrial setting. Our findings indicate that while phenotypic data and biomedical imaging present complex challenges, machine learning techniques can play a pivotal role in making early drug discovery more efficient and effective.
|
|
| 4. |
- Fredin Haslum, Johan, et al.
(författare)
-
Metadata-guided Consistency Learning for High Content Images
- 2023
-
Ingår i: PLMR: Volume 227: Medical Imaging with Deep Learning, 10-12 July 2023, Nashville, TN, USA.
-
Konferensbidrag (refereegranskat)abstract
- High content imaging assays can capture rich phenotypic response data for large sets of compound treatments, aiding in the characterization and discovery of novel drugs. However, extracting representative features from high content images that can capture subtle nuances in phenotypes remains challenging. The lack of high-quality labels makes it difficult to achieve satisfactory results with supervised deep learning. Self-Supervised learning methods have shown great success on natural images, and offer an attractive alternative also to microscopy images. However, we find that self-supervised learning techniques underperform on high content imaging assays. One challenge is the undesirable domain shifts present in the data known as batch effects, which are caused by biological noise or uncontrolled experimental conditions. To this end, we introduce Cross-Domain Consistency Learning (CDCL), a self-supervised approach that is able to learn in the presence of batch effects. CDCL enforces the learning of biological similarities while disregarding undesirable batch-specific signals, leading to more useful and versatile representations. These features are organised according to their morphological changes and are more useful for downstream tasks – such as distinguishing treatments and mechanism of action.
|
|
| 5. |
- Jarvius, Malin, et al.
(författare)
-
In situ detection of phosphorylated platelet-derived growth factor receptor beta using a generalized proximity ligation method
- 2007
-
Ingår i: Molecular & Cellular Proteomics. - 1535-9476 .- 1535-9484. ; 6:9, s. 1500-1509
-
Tidskriftsartikel (refereegranskat)abstract
- Improved methods are needed for in situ characterization of post-translational modifications in cell lines and tissues. For example, it is desirable to monitor the phosphorylation status of individual receptor tyrosine kinases in samples from human tumors treated with inhibitors to evaluate therapeutic responses. Unfortunately the leading methods for observing the dynamics of tissue post-translational modifications in situ, immunohistochemistry and immunofluorescence, exhibit limited sensitivity and selectivity. Proximity ligation assay is a novel method that offers improved selectivity through the requirement of dual recognition and increased sensitivity by including DNA amplification as a component of detection of the target molecule. Here we therefore established a generalized in situ proximity ligation assay to investigate phosphorylation of platelet-derived growth factor receptor β (PDGFRβ) in cells stimulated with platelet-derived growth factor BB. Antibodies specific for immunoglobulins from different species, modified by attachment of DNA strands, were used as secondary proximity probes together with a pair of primary antibodies from the corresponding species. Dual recognition of receptors and phosphorylated sites by the primary antibodies in combination with the secondary proximity probes was used to generate circular DNA strands; this was followed by signal amplification by replicating the DNA circles via rolling circle amplification. We detected tyrosine phosphorylated PDGFRβ in human embryonic kidney cells stably overexpressing human influenza hemagglutinin-tagged human PDGFRβ in porcine aortic endothelial cells transfected with the β-receptor, but not in cells transfected with the α-receptor, and also in immortalized human foreskin fibroblasts, BJ hTert, endogenously expressing the PDGFRβ. We furthermore visualized tyrosine phosphorylated PDGFRβ in tissue sections from fresh frozen human scar tissue undergoing wound healing. The method should be of great value to study signal transduction, screen for effects of pharmacological agents, and enhance the diagnostic potential in histopathology.
|
|
| 6. |
- Leuchowius, Karl-Johan, et al.
(författare)
-
Parallel Visualization of Multiple Protein Complexes in Individual Cells in Tumor Tissue
- 2013
-
Ingår i: Molecular & Cellular Proteomics. - 1535-9476 .- 1535-9484. ; 12:6, s. 1563-1571
-
Tidskriftsartikel (refereegranskat)abstract
- Cellular functions are regulated and executed by complex protein interaction networks. Accordingly, it is essential to understand the interplay between proteins in determining the activity status of signaling cascades. New methods are therefore required to provide information on different protein interaction events at the single cell level in heterogeneous cell populations such as in tissue sections. Here, we describe a multiplex proximity ligation assay for simultaneous visualization of multiple protein complexes in situ. The assay is an enhancement of the original proximity ligation assay, and it is based on using proximity probes labeled with unique tag sequences that can be used to read out which probes, from a pool of probes, have bound a certain protein complex. Using this approach, it is possible to gain information on the constituents of different protein complexes, the subcellular location of the complexes, and how the balance between different complex constituents can change between normal and malignant cells, for example. As a proof of concept, we used the assay to simultaneously visualize multiple protein complexes involving EGFR, HER2, and HER3 homo- and heterodimers on a single-cell level in breast cancer tissue sections. The ability to study several protein complex formations concurrently at single cell resolution could be of great potential for a systems understanding, paving the way for improved disease diagnostics and possibilities for drug development.
|
|
| 7. |
|
|
| 8. |
- Boström, Hans, et al.
(författare)
-
U-2973, a novel B-cell line established from a patient with a mature B-cell leukemia displaying concurrent t(14;18) and MYC translocation to a non-IG gene partner
- 2008
-
Ingår i: European Journal of Haematology. - : Wiley. - 0902-4441 .- 1600-0609. ; 81:3, s. 218-225
-
Tidskriftsartikel (refereegranskat)abstract
- B-cell lymphomas/leukemias with simultaneous t(14;18)(q32;q21) and MYC rearrangements have recently been shown to constitute a separate diagnostic entity, presenting with a rapid clinical course and a very poor prognosis. We describe the establishment of an Epstein-Barr virus negative cell line, designated U-2973, from a male patient with a de novo aggressive B-cell lymphoma/leukemia and very high peripheral blast cell count. Flow cytometry of bone marrow cells and U-2973 displayed a mature B-cell phenotype, and immunostaining showed expression of MYC and BCL2. IG gene rearrangement data were consistent with a lymphoid neoplasm of germinal centre derivation. Cytogenetic studies using conventional G-banding, fluorescent in situ hybridization, spectral karyotyping and single nucleotide polymorphism array demonstrated a complex karyotype with both a t(14;18) and double translocations between MYC and a non-IG gene partner located at chromosome 12p12.1.
|
|
| 9. |
- Cane, Gaëlle, et al.
(författare)
-
Protein Diagnostics by Proximity Ligation: Combining Multiple Recognition and DNA Amplification for Improved Protein Analyses
- 2017. - 3
-
Ingår i: Molecular Diagnostics (Third Edition). - 2016 : Academia Press. - 9780128029718 ; , s. 219-231
-
Bokkapitel (refereegranskat)abstract
- Proximity ligation assay (PLA) is a unique method in which single-stranded oligonucleotides are conjugated to affinity binders of proteins, followed by amplification of the signal by DNA polymerization and hybridization of complementary oligonucleotides labeled with fluorogenic or chromogenic readout. Here, a brief overview of the field of protein analysis describes the background and the initial development of the technique for the detection of protein–protein interactions via the proximity probes mentioned. In this context, PLA can constrain the general problem of cross-reactivity in protein detection by affinity binders, by ensuring that only cognate pairs of proximity probes result in a signal. Thereafter, this chapter deals mainly with derivatives methods and their applications, with a particular interest in improved specificity, application to various biological materials, and multiplexing. The method has been applied in situ and in solution, adapted for the detection of posttranslational modifications such as phosphorylation and interactions between proteins and specific DNA sequences, and multiplexed to a certain extent, which illustrates its versatility. A technique free from enzymatic reaction, the hybridization chain reaction, can be considered a cost-effective alternative particularly suitable to molecular diagnostics. Finally, we explore further development toward higher-level multiplexing and sensitivity. At this point it is not clear what level can be achieved by PLA, but the assay is compatible with a wide range of readout, including separate real-time amplification reactions and novel microfluidic read-out platforms.
|
|
| 10. |
- Johansson Wensman, Jonas, et al.
(författare)
-
Visualization of Borna Disease Virus Protein Interactions with Host Proteins using in situ Proximity Ligation Assay
- 2016
-
Ingår i: British journal of virology. - : ResearchersLinks Ltd. - 2055-6128. ; 3:1, s. 11-23
-
Tidskriftsartikel (refereegranskat)abstract
- Borna disease virus type 1 (BDV) comprises highly conserved neurotropic non-segmented negative strand RNA-virus variants causing neurological and behavioral disorders in a wide range of mammalian animals, possibly including humans. Viral persistence in the brain has been frequently observed, however, the exact mechanisms behind BDV’s ability to establish persistence despite a prominent immune response are not known. Here we have used in situ proximity ligation assay (in situ PLA), a selective tool for studying virus-host protein-protein interactions. BDV P (phosphoprotein) and N (nucleoprotein) have previously been reported to interact with several host proteins, thereby interfering with various signaling pathways. In this study, we focused on some of these interactions (BDV P-HMGB1, BDV N/P-Cdc2). First, we used rat glioma cell cultures persistently infected with a laboratory strain of BDV (C6BV) to establish the assay. Next, in situ PLA was applied to detect BDV P in brain tissues of infected animals. Finally, protein-protein interactions were visualized in both C6BV and brain tissues of experimentally as well as naturally infected animals (rat and horse, respectively). BDV proteins and their interactions with host proteins could be shown in cell cultures (HMGB1, Cdc2) and in brain tissues of rat (HMGB1, Cdc2) and horse (Cdc2 only) infected with BDV. In this study, we have for the first time directly visualized protein-protein interactions between BDV and its host, and thereby confirmed previous data to demonstrate findings in cell cultures to be applicable also in experimentally and naturally infected animals.
|
|
