1
|
Michael R, Kæstel-Hansen J, Mørch Groth P, Bartels S, Salomon J, Tian P, Hatzakis NS, Boomsma W. A systematic analysis of regression models for protein engineering. PLoS Comput Biol 2024; 20:e1012061. [PMID: 38701099 DOI: 10.1371/journal.pcbi.1012061] [Citation(s) in RCA: 0] [Impact Index Per Article: 0] [Reference Citation Analysis] [What about the content of this article? (0)] [Affiliation(s)] [Abstract] [Track Full Text] [Journal Information] [Subscribe] [Scholar Register] [Received: 09/26/2023] [Accepted: 04/10/2024] [Indexed: 05/05/2024] Open
Abstract
To optimize proteins for particular traits holds great promise for industrial and pharmaceutical purposes. Machine Learning is increasingly applied in this field to predict properties of proteins, thereby guiding the experimental optimization process. A natural question is: How much progress are we making with such predictions, and how important is the choice of regressor and representation? In this paper, we demonstrate that different assessment criteria for regressor performance can lead to dramatically different conclusions, depending on the choice of metric, and how one defines generalization. We highlight the fundamental issues of sample bias in typical regression scenarios and how this can lead to misleading conclusions about regressor performance. Finally, we make the case for the importance of calibrated uncertainty in this domain.
Collapse
Affiliation(s)
- Richard Michael
- Department of Computer Science, University of Copenhagen, Copenhagen, Denmark
| | | | - Peter Mørch Groth
- Department of Computer Science, University of Copenhagen, Copenhagen, Denmark
- Enzyme Research, Novonesis, Kongens Lyngby, Denmark
| | - Simon Bartels
- Department of Computer Science, University of Copenhagen, Copenhagen, Denmark
| | | | - Pengfei Tian
- Enzyme Research, Novozymes A/S, Kongens Lyngby, Denmark
| | - Nikos S Hatzakis
- Department of Chemistry, University of Copenhagen, Copenhagen, Denmark
| | - Wouter Boomsma
- Department of Computer Science, University of Copenhagen, Copenhagen, Denmark
| |
Collapse
|
2
|
Bender SWB, Dreisler MW, Zhang M, Kæstel-Hansen J, Hatzakis NS. SEMORE: SEgmentation and MORphological fingErprinting by machine learning automates super-resolution data analysis. Nat Commun 2024; 15:1763. [PMID: 38409214 PMCID: PMC10897458 DOI: 10.1038/s41467-024-46106-0] [Citation(s) in RCA: 0] [Impact Index Per Article: 0] [Reference Citation Analysis] [What about the content of this article? (0)] [Affiliation(s)] [Abstract] [Key Words] [MESH Headings] [Grants] [Track Full Text] [Journal Information] [Subscribe] [Scholar Register] [Received: 05/10/2023] [Accepted: 02/13/2024] [Indexed: 02/28/2024] Open
Abstract
The morphology of protein assemblies impacts their behaviour and contributes to beneficial and aberrant cellular responses. While single-molecule localization microscopy provides the required spatial resolution to investigate these assemblies, the lack of universal robust analytical tools to extract and quantify underlying structures limits this powerful technique. Here we present SEMORE, a semi-automatic machine learning framework for universal, system- and input-dependent, analysis of super-resolution data. SEMORE implements a multi-layered density-based clustering module to dissect biological assemblies and a morphology fingerprinting module for quantification by multiple geometric and kinetics-based descriptors. We demonstrate SEMORE on simulations and diverse raw super-resolution data: time-resolved insulin aggregates, and published data of dSTORM imaging of nuclear pore complexes, fibroblast growth receptor 1, sptPALM of Syntaxin 1a and dynamic live-cell PALM of ryanodine receptors. SEMORE extracts and quantifies all protein assemblies, their temporal morphology evolution and provides quantitative insights, e.g. classification of heterogeneous insulin aggregation pathways and NPC geometry in minutes. SEMORE is a general analysis platform for super-resolution data, and being a time-aware framework can also support the rise of 4D super-resolution data.
Collapse
Affiliation(s)
- Steen W B Bender
- Department of Chemistry, University of Copenhagen, Copenhagen, Denmark
- Center for 4D cellular dynamics, University of Copenhagen, Copenhagen, Denmark
- Novo Nordisk Center for Optimised Oligo Escape and Control of Disease, University of Copenhagen, Copenhagen, Denmark
| | - Marcus W Dreisler
- Department of Chemistry, University of Copenhagen, Copenhagen, Denmark
- Center for 4D cellular dynamics, University of Copenhagen, Copenhagen, Denmark
- Novo Nordisk Center for Optimised Oligo Escape and Control of Disease, University of Copenhagen, Copenhagen, Denmark
| | - Min Zhang
- Department of Chemistry, University of Copenhagen, Copenhagen, Denmark
- Center for 4D cellular dynamics, University of Copenhagen, Copenhagen, Denmark
- Novo Nordisk Center for Optimised Oligo Escape and Control of Disease, University of Copenhagen, Copenhagen, Denmark
| | - Jacob Kæstel-Hansen
- Department of Chemistry, University of Copenhagen, Copenhagen, Denmark.
- Center for 4D cellular dynamics, University of Copenhagen, Copenhagen, Denmark.
- Novo Nordisk Center for Optimised Oligo Escape and Control of Disease, University of Copenhagen, Copenhagen, Denmark.
| | - Nikos S Hatzakis
- Department of Chemistry, University of Copenhagen, Copenhagen, Denmark.
- Center for 4D cellular dynamics, University of Copenhagen, Copenhagen, Denmark.
- Novo Nordisk Center for Optimised Oligo Escape and Control of Disease, University of Copenhagen, Copenhagen, Denmark.
- Novo Nordisk Center for Protein Research, University of Copenhagen, Copenhagen, Denmark.
| |
Collapse
|
3
|
Kæstel-Hansen J, Hatzakis NS. Everything, everywhere, almost at once. eLife 2024; 13:e95362. [PMID: 38285015 PMCID: PMC10824506 DOI: 10.7554/elife.95362] [Citation(s) in RCA: 0] [Impact Index Per Article: 0] [Reference Citation Analysis] [What about the content of this article? (0)] [Affiliation(s)] [Abstract] [Key Words] [MESH Headings] [Track Full Text] [Journal Information] [Subscribe] [Scholar Register] [Indexed: 01/30/2024] Open
Abstract
A new platform that can follow the movement of individual proteins inside millions of cells in a single day will help contribute to existing knowledge of cell biology and identify new therapeutics.
Collapse
Affiliation(s)
- Jacob Kæstel-Hansen
- Department of Chemistry, Novo Nordisk Foundation Center for Optimized Oligo Escape and Control of Disease, University of CopenhagenCopenhagenDenmark
| | - Nikos S Hatzakis
- Department of Chemistry, Novo Nordisk Foundation Center for Optimized Oligo Escape and Control of Disease, University of CopenhagenCopenhagenDenmark
| |
Collapse
|
4
|
Kæstel-Hansen J, de Sautu M, Saminathan A, Scanavachi G, Da Cunha Correia RFB, Nielsen AJ, Bleshøy SV, Boomsma W, Kirchhausen T, Hatzakis NS. Deep learning assisted single particle tracking for automated correlation between diffusion and function. bioRxiv 2023:2023.11.16.567393. [PMID: 38014323 PMCID: PMC10680793 DOI: 10.1101/2023.11.16.567393] [Citation(s) in RCA: 0] [Impact Index Per Article: 0] [Reference Citation Analysis] [What about the content of this article? (0)] [Affiliation(s)] [Abstract] [Grants] [Track Full Text] [Subscribe] [Scholar Register] [Indexed: 11/29/2023]
Abstract
Sub-cellular diffusion in living systems reflects cellular processes and interactions. Recent advances in optical microscopy allow the tracking of this nanoscale diffusion of individual objects with an unprecedented level of precision. However, the agnostic and automated extraction of functional information from the diffusion of molecules and organelles within the sub-cellular environment, is labor-intensive and poses a significant challenge. Here we introduce DeepSPT, a deep learning framework to interpret the diffusional 2D or 3D temporal behavior of objects in a rapid and efficient manner, agnostically. Demonstrating its versatility, we have applied DeepSPT to automated mapping of the early events of viral infections, identifying distinct types of endosomal organelles, and clathrin-coated pits and vesicles with up to 95% accuracy and within seconds instead of weeks. The fact that DeepSPT effectively extracts biological information from diffusion alone indicates that besides structure, motion encodes function at the molecular and subcellular level.
Collapse
Affiliation(s)
- Jacob Kæstel-Hansen
- Department of Chemistry University of Copenhagen
- Center for 4D cellular dynamics, Department of Chemistry University of Copenhagen
- Novo Nordisk Center for Optimised Oligo Escape
- Novo Nordisk foundation Center for Protein Research
| | - Marilina de Sautu
- Biological Chemistry and Molecular Pharmaceutics Harvard Medical School
- Laboratory of Molecular Medicine Boston Children's Hospital
| | - Anand Saminathan
- Department of Cell Biology Harvard Medical School
- Department of Pediatrics Harvard Medical School
- Program in Cellular and Molecular Medicine Boston Children's Hospital
| | - Gustavo Scanavachi
- Department of Cell Biology Harvard Medical School
- Department of Pediatrics Harvard Medical School
- Program in Cellular and Molecular Medicine Boston Children's Hospital
| | - Ricardo F Bango Da Cunha Correia
- Department of Cell Biology Harvard Medical School
- Department of Pediatrics Harvard Medical School
- Program in Cellular and Molecular Medicine Boston Children's Hospital
| | - Annette Juma Nielsen
- Department of Chemistry University of Copenhagen
- Center for 4D cellular dynamics, Department of Chemistry University of Copenhagen
- Novo Nordisk Center for Optimised Oligo Escape
- Novo Nordisk foundation Center for Protein Research
| | - Sara Vogt Bleshøy
- Department of Chemistry University of Copenhagen
- Center for 4D cellular dynamics, Department of Chemistry University of Copenhagen
- Novo Nordisk Center for Optimised Oligo Escape
- Novo Nordisk foundation Center for Protein Research
| | | | - Tom Kirchhausen
- Department of Cell Biology Harvard Medical School
- Department of Pediatrics Harvard Medical School
- Program in Cellular and Molecular Medicine Boston Children's Hospital
| | - Nikos S Hatzakis
- Department of Chemistry University of Copenhagen
- Center for 4D cellular dynamics, Department of Chemistry University of Copenhagen
- Novo Nordisk Center for Optimised Oligo Escape
- Novo Nordisk foundation Center for Protein Research
| |
Collapse
|
5
|
Benning NA, Kæstel-Hansen J, Rashid F, Park S, Merino Urteaga R, Liao TW, Hao J, Berger JM, Hatzakis NS, Ha T. Dimensional Reduction for Single-Molecule Imaging of DNA and Nucleosome Condensation by Polyamines, HP1α and Ki-67. J Phys Chem B 2023; 127:1922-1931. [PMID: 36853329 PMCID: PMC10009747 DOI: 10.1021/acs.jpcb.2c07011] [Citation(s) in RCA: 0] [Impact Index Per Article: 0] [Reference Citation Analysis] [What about the content of this article? (0)] [Affiliation(s)] [Abstract] [Track Full Text] [Journal Information] [Subscribe] [Scholar Register] [Indexed: 03/01/2023]
Abstract
Macromolecules organize themselves into discrete membrane-less compartments. Mounting evidence has suggested that nucleosomes as well as DNA itself can undergo clustering or condensation to regulate genomic activity. Current in vitro condensation studies provide insight into the physical properties of condensates, such as surface tension and diffusion. However, methods that provide the resolution needed for complex kinetic studies of multicomponent condensation are desired. Here, we use a supported lipid bilayer platform in tandem with total internal reflection microscopy to observe the two-dimensional movement of DNA and nucleosomes at the single-molecule resolution. This dimensional reduction from three-dimensional studies allows us to observe the initial condensation events and dissolution of these early condensates in the presence of physiological condensing agents. Using polyamines, we observed that the initial condensation happens on a time scale of minutes while dissolution occurs within seconds upon charge inversion. Polyamine valency, DNA length, and GC content affect the threshold polyamine concentration for condensation. Protein-based nucleosome condensing agents, HP1α and Ki-67, have much lower threshold concentrations for condensation than charge-based condensing agents, with Ki-67 being the most effective, requiring as low as 100 pM for nucleosome condensation. In addition, we did not observe condensate dissolution even at the highest concentrations of HP1α and Ki-67 tested. We also introduce a two-color imaging scheme where nucleosomes of high density labeled in one color are used to demarcate condensate boundaries and identical nucleosomes of another color at low density can be tracked relative to the boundaries after Ki-67-mediated condensation. Our platform should enable the ultimate resolution of single molecules in condensation dynamics studies of chromatin components under defined physicochemical conditions.
Collapse
Affiliation(s)
- Nils A Benning
- Department of Biology, Johns Hopkins University, Baltimore, Maryland 21218, United States
| | - Jacob Kæstel-Hansen
- Department of Chemistry and Nanoscience Centre, University of Copenhagen, Copenhagen 2100, Denmark
| | - Fahad Rashid
- Department of Biophysics and Biophysical Chemistry, Johns Hopkins University School of Medicine, Baltimore, Maryland 21205, United States
| | - Sangwoo Park
- Department of Biophysics and Biophysical Chemistry, Johns Hopkins University School of Medicine, Baltimore, Maryland 21205, United States
| | - Raquel Merino Urteaga
- Department of Biology, Johns Hopkins University, Baltimore, Maryland 21218, United States
| | - Ting-Wei Liao
- Department of Biophysics, Johns Hopkins University, Baltimore, Maryland 21218, United States
| | - Jingzhou Hao
- Department of Biophysics, Johns Hopkins University, Baltimore, Maryland 21218, United States
| | - James M Berger
- Department of Biophysics and Biophysical Chemistry, Johns Hopkins University School of Medicine, Baltimore, Maryland 21205, United States
| | - Nikos S Hatzakis
- Department of Chemistry and Nanoscience Centre, University of Copenhagen, Copenhagen 2100, Denmark.,Novo Nordisk Foundation Centre for Protein Research, Faculty of Health and Medical Sciences, University of Copenhagen, Copenhagen 2100, Denmark
| | - Taekjip Ha
- Department of Biology, Johns Hopkins University, Baltimore, Maryland 21218, United States.,Department of Biophysics and Biophysical Chemistry, Johns Hopkins University School of Medicine, Baltimore, Maryland 21205, United States.,Department of Biophysics, Johns Hopkins University, Baltimore, Maryland 21218, United States.,Department of Biomedical Engineering, Johns Hopkins University, Baltimore, Maryland 21218, United States.,Howard Hughes Medical Institute, Baltimore, Maryland 21205, United States
| |
Collapse
|
6
|
Kæstel-Hansen J, Kreutzberger A, Kirchhausen T, Hatzakis NS. Rapid extraction of time-dependent behaviors of endo-lysosomal structures and their cargo aided by deep learning. Biophys J 2023; 122:517a. [PMID: 36784675 DOI: 10.1016/j.bpj.2022.11.2751] [Citation(s) in RCA: 0] [Impact Index Per Article: 0] [Reference Citation Analysis] [What about the content of this article? (0)] [Affiliation(s)] [Track Full Text] [Journal Information] [Subscribe] [Scholar Register] [Indexed: 02/12/2023] Open
Affiliation(s)
| | | | | | - Nikos S Hatzakis
- Nano-Science Center, University of Copenhagen, Copenhagen, Denmark
| |
Collapse
|
7
|
Bender S, Kæstel-Hansen J, Zhang M, Hatzakis NS. Direct observation and classification of heterogeneous protein aggregation in real-time through super-resolution and aggregational fingerprinting. Biophys J 2023; 122:16a. [PMID: 36782796 DOI: 10.1016/j.bpj.2022.11.315] [Citation(s) in RCA: 0] [Impact Index Per Article: 0] [Reference Citation Analysis] [What about the content of this article? (0)] [Affiliation(s)] [Track Full Text] [Journal Information] [Subscribe] [Scholar Register] [Indexed: 02/12/2023] Open
Affiliation(s)
- Steen Bender
- Department of Chemistry, University of Copenhagen, Copenhagen, Denmark
| | | | - Min Zhang
- University of Copenhagen, Frederiksberg Copenhagen, Denmark
| | - Nikos S Hatzakis
- Nano-Science Center, University of Copenhagen, Copenhagen, Denmark; Department of Chemistry, Center for 4D Cellular Dynamics, Copenhagen, Denmark
| |
Collapse
|
8
|
Kæstel-Hansen J, Bohr SSR, Høgh Schulz F, Juma Nielsen A, Krogh Boomsma W, Hatzakis NS. Accelerating single particle discoveries using machine learning. Biophys J 2022. [DOI: 10.1016/j.bpj.2021.11.1906] [Citation(s) in RCA: 0] [Impact Index Per Article: 0] [Reference Citation Analysis] [What about the content of this article? (0)] [Track Full Text] [Journal Information] [Subscribe] [Scholar Register] [Indexed: 11/29/2022] Open
|
9
|
Moses ME, Lund PM, Bohr SSR, Iversen JF, Kæstel-Hansen J, Kallenbach AS, Iversen L, Christensen SM, Hatzakis NS. Single-Molecule Study of Thermomyces lanuginosus Lipase in a Detergency Application System Reveals Diffusion Pattern Remodeling by Surfactants and Calcium. ACS Appl Mater Interfaces 2021; 13:33704-33712. [PMID: 34235926 DOI: 10.1021/acsami.1c08809] [Citation(s) in RCA: 5] [Impact Index Per Article: 1.7] [Reference Citation Analysis] [What about the content of this article? (0)] [Affiliation(s)] [Abstract] [Key Words] [MESH Headings] [Track Full Text] [Subscribe] [Scholar Register] [Indexed: 06/13/2023]
Abstract
Lipases comprise one of the major enzyme classes in biotechnology with applications within, e.g., baking, brewing, biocatalysis, and the detergent industry. Understanding the mechanisms of lipase function and regulation is therefore important to facilitate the optimization of their function by protein engineering. Advances in single-molecule studies in model systems have provided deep mechanistic insights on lipase function, such as the existence of functional states, their dependence on regulatory cues, and their correlation to activity. However, it is unclear how these observations translate to enzyme behavior in applied settings. Here, single-molecule tracking of individual Thermomyces lanuginosus lipase (TLL) enzymes in a detergency application system allowed real-time direct observation of spatiotemporal localization, and thus diffusional behavior, of TLL enzymes on a lard substrate. Parallelized imaging of thousands of individual enzymes allowed us to observe directly the existence and quantify the abundance and interconversion kinetics between three diffusional states that we recently provided evidence to correlate with function. We observe redistribution of the enzyme's diffusional pattern at the lipid-water interface as well as variations in binding efficiency in response to surfactants and calcium, demonstrating that detergency effectors can drive the sampling of lipase functional states. Our single-molecule results combined with ensemble activity assays and enzyme surface binding efficiency readouts allowed us to deconvolute how application conditions can significantly alter protein functional dynamics and/or surface binding, both of which underpin enzyme performance. We anticipate that our results will inspire further efforts to decipher and integrate the dynamic nature of lipases, and other enzymes, in the design of new biotechnological solutions.
Collapse
Affiliation(s)
- Matias E Moses
- Novozymes A/S, Biologiens Vej 2, DK-2800 Kgs. Lyngby, Denmark
- Department of Chemistry & Nano-science Center, University of Copenhagen, Thorvaldsensvej 40, DK-1871 Frederiksberg C, Denmark
| | - Philip M Lund
- Novozymes A/S, Biologiens Vej 2, DK-2800 Kgs. Lyngby, Denmark
- Department of Chemistry & Nano-science Center, University of Copenhagen, Thorvaldsensvej 40, DK-1871 Frederiksberg C, Denmark
| | - Søren S-R Bohr
- Department of Chemistry & Nano-science Center, University of Copenhagen, Thorvaldsensvej 40, DK-1871 Frederiksberg C, Denmark
| | - Josephine F Iversen
- Department of Chemistry & Nano-science Center, University of Copenhagen, Thorvaldsensvej 40, DK-1871 Frederiksberg C, Denmark
| | - Jacob Kæstel-Hansen
- Department of Chemistry & Nano-science Center, University of Copenhagen, Thorvaldsensvej 40, DK-1871 Frederiksberg C, Denmark
| | - Amalie S Kallenbach
- Department of Chemistry & Nano-science Center, University of Copenhagen, Thorvaldsensvej 40, DK-1871 Frederiksberg C, Denmark
| | - Lars Iversen
- Novozymes A/S, Biologiens Vej 2, DK-2800 Kgs. Lyngby, Denmark
| | | | - Nikos S Hatzakis
- Department of Chemistry & Nano-science Center, University of Copenhagen, Thorvaldsensvej 40, DK-1871 Frederiksberg C, Denmark
- Novo Nordisk Foundation Centre for Protein Research, Faculty of Health and Medical Sciences, University of Copenhagen, Blegdamsvej 3B, DK-2200 Copenhagen N, Denmark
| |
Collapse
|