Progress 10/01/19 to 09/30/20
Outputs
Target Audience:Target audiences have included both the plant science and applied mathematics communities. The proposed research seeks to measure plant morphology in exquisite detail using X-ray CT and to apply Topological Data Analysis (TDA) to extract the information embedded within phenotype comprehensively and to more fully understand the genetic and environmental basis of traits. The above requires not only working closely with plant scientists but communicating their needs to applied mathematicians so that an underlying theory can be developed. Plant scientists and applied mathematician audiences have both been engaged during the reporting period, through collaboration, publications, and attending meetings. In the first years of the project there was a more quantitative focus to get the X-ray CT up and running, an image analysis pipeline for handling the large amounts of data being generated, and exploring the foundational computational mathematics required to statistically analyze the data. The above is still an area of active work, but especially in the last reporting period, we are working more closely with biologists to begin asking fundamental scientific questions and working towards proposals. The main focus at the moment is using Topological Data Analysis (TDA) to not only quantify the overall morphology of time series of plant growth, but of underlying time series of gene expression networks as well. Using the resulting topological signatures to predict phenotype from genotype and vice versa, the target audience is biology in general, for which this is a grand challenge, to demonstrate the usefulness of mathematics in solving pressing problems in the plant sciences. Students are another target audience that have been reached through both formal classroom instruction and the development of innovative teaching methodologies. Educating plant science students that are fluent in coding and mathematics is necessary to carry out the proposed research but finding students of the required proficiency is difficult. Previously, we developed a course that teaches Python and mathematical modeling to students who do not necessarily have experience in either. The course was designed in a way that was prepared for the global pandemic. Because all code was posted online using interactive Jupyter notebooks, over four hours of lecture were publicly posted, and a flipped classroom design was used, there was no problem making the course virtual. Beyond MSU students, a few international students are the target audience as well, who have taken advantage of the public materials and taken the course. The innovative methods developed for course are currently being written up as educational research to share with other lecturers, which we hope will be impactful in addressing the challenges of the pandemic to education in general. Changes/Problems:There are no significant changes or problems. Dr. Tim Ophelders has left, graduate student Kayla Makela has started, and two post-doc offers have been made. What opportunities for training and professional development has the project provided?Post-doc/X-ray CT manager Dr. Michelle Quigley (70/30 Chitwood/Munch): Dr. Quigley has become proficient in X-ray imaging and manages the X-ray CT machine. She has become versed in the safety and operating protocols for the machine. She will be focusing more developing databases to disseminate and manage X-ray CT online. Post-doc Dr. Tim Ophelders (30/70 Chitwood/Munch): Dr. Ophelders has mentored graduate students and undergraduates in computation and mathematical analyses of data. He has attended professional meetings related to Topological Data Analysis (TDA). Dr. Ophelders left the program in Summer 2020. Graduate student Kayla Makela (50/50 Chitwood/Munch): Kayla Makela started as a graduate student using X-ray CT to study tree rings and climate prediction, in collaboration with Dr. Asia Dowtin in Forestry. Graduate student Erik Amézquita (50/50 Chitwood/Munch): Erik Amézquita has attended several mathematical conferences and taken classes in both mathematics and plant biology to improve his skills applying TDA to the analysis of plant morphology. Erik has been instrumental in developing image anlaysis and TDA techniques for the barley project. Undergraduate Joey Mullins (100 Chitwood): Joey is part of the MSU Professorial Assistants program. Joey has learned how to operate the X-ray CT machine and assists Dr. Michelle Quigley in scanning projects. How have the results been disseminated to communities of interest?The course co-developed by Dr. Chitwood and Dr. VanBuren to teach Python to those with no prior coding experience continues to be improved. The course focuses on teaching mathematical modeling of plant morphology and bioinformatics. The course will help train plant biologists versed in coding to tackle the Objectives outlined in this proposal. The course has been made public to disseminate these materials broadly and its content is being written up as educational research. Additionally, from the first year a product is a manuscript that was recently accepted with all students as co-authors. The writing up of a class manuscript that will be published will continue to be a class objective, to help students experience the publication process and disseminate results. What do you plan to do during the next reporting period to accomplish the goals?For the next reporting period, an emphasis will be placed on recruiting personnel and especially an emphasis on obtaining funding for phenotypic and genotypic time series and predicting phenotype from molecular profiles using Topological Data Analysis. Two post-doc offers have been recently made to increase personnel (and some delays because of immigration problems caused by Presidential executive orders) and writing for a proposal has begun as well.
Impacts
What was accomplished under these goals?
In a world of big data, in which we can measure vast quantities of information quickly, the primary problem we encounter is how to make sense of what we have measured. We think of data in terms of human-generated information, but the natural world contains multitudes more data. The form of every organism is exquisitely encoded by genetic information and sculpted by environmental forces. Evolutionary forces are responsible for feeding the world's civilizations through domestication, and environmental responses will be the key to mitigating the effects of climate change on our crops. If we could not only record everything that a plant is, but also quantify that information, so that we could make predictions using genetic and environmental variables, we would be in a better place to rationally-design the plants and agriculture of the future. The proposed work uses a combination of X-ray Computed Tomography (CT) and a mathematical field known as Topological Data Analysis (TDA) to record and measure the plant form comprehensively. We use industrial X-ray CT (just like the medical technology) to create 3D reconstructions of plants--both internally and externally--at extremely high resolution. We use mathematics to quantify the complex morphologies of plants, considering the entirety of data, from the organ to tissue levels in multiple dimensions. The resulting information, a measurement of what a plant is, is then modeled with genetic and environmental information. Below, some applications of the above work relating to the proposed major goals are outlined: Objective 1: Analyze 3D objects using a variety of filter functions, and compare the outputs using benchmarks such as heritability and predictability. 1.Experiments conducted: During the previous research period, we have started working with numerous plant science collaborators and scanned large datasets using X-ray CT, including: 1) tracking the effects of selection on barley grain morphology in a long-term evolution experiment, 2) studying the evolution of Citrus morphology during evolution and in hybridization events, 3) creating time-lapses of Arabidopsis and succulent growth and integrating these with gene expression dataset, 4) predicting shelling properties of walnuts from their shape, 5) reconstructing the domestication history of sunflower using ancient preserved discs and DNA, and 6) measuring tree rings with X-ray CT to measure climate. 2.Data collected: Currently, we are focusing our efforts in analyzing the barley dataset. Using image analysis techniques, we have isolated over 40,000 barley seeds from a long-term artificial evolution experiment over 60 generations. With collaborators, these barley accessions have been genetically sequenced. 3.Summary Statistics and summary of results: Using the Euler Characteristic Transform (ECT), the shape of each of 40,000 barley seeds has been quantified. For each group of barley founders and subsequent progeny generations, the overall similarity and shape differences between lines phenotypically can be compared. 4.Key outcomes: The extracted shape information can be used for genetic analyses, as well as population genetic analysis, which shows the dominance of genetic material from some founders over others across evolution. Combining the phenotypic and genetic results to see how these genetic changes have effected morphological outcomes is on-going with collaborators. Objective 2: Create a framework to analyze dynamic 4D objects through time. 1.Experiments conducted: A key objective of the proposed work is to not only study static plant shapes, but dynamic ones as well, especially plant development. Another objective is to more generally use Topological Data Analysis (TDA) to describe the shape of other datasets, such as gene expression. 2.Data collected: We have taken X-ray CT time-series of growing plants and collected tissues for RNA-Seq analysis. 3.Summary Statistics and summary of results: Several succulents and two Arabidopsis accessions with contrasting leaf shapes and developmental stability were imaged as a time series and tissue collected for molecular analysis. 4.Key outcomes: The analysis of this data (as for last reporting period) is still in process. However, two offer letters for post-docs to work on this project have been made and this project will be the source of a proposal. Therefore, this Objective is a major focus for the next reporting period. By measuring phenotype as a time series at both the phenotypic and molecular levels, phenotype can be predicted from molecular profiles, a Grand Challenge in biology. Objective 3: Explore Topological Data Analysis (TDA) methods to visualize, interpret, and disseminate X-ray CT data and morphological analyses. 1.Experiments conducted: As described above in the previous two objectives, large amounts of X-ray CT data are being generated. The dissemination of this data to the public remains a challenge. 2.Data collected: Hundreds of large X-ray CT datasets. 3.Summary Statistics and summary of results: A server is being set-up to store the data. Recently, an NSF award " 4.Key outcomes: In the coming years, we will be using advanced cyberinfrastructure to share large X-ray CT datasets and analysis methods. The above should foster are larger and more participatory community in regards to using Topological Data Analysis methods in the plant sciences.
Publications
- Type:
Journal Articles
Status:
Published
Year Published:
2019
Citation:
Li M, Klein LL. Duncan KE, Jiang N, Chitwood DH, Londo JP, Miller A, Topp C (2019) Characterizing grapevine (Vitis spp.) inflorescence architecture using X-ray imaging: implications for understanding cluster density. J Exp Botany. 70(21):6261-6276.. DOI: https://doi.org/10.1093/jxb/erz394. Pre-print: https://doi.org/10.1101/557819
- Type:
Journal Articles
Status:
Published
Year Published:
2019
Citation:
Demmings (Takacs) EM, Williams B, Lee CR, Burgos PB, Hwang CF, Reisch BI, Chitwood DH, Londo JP (2019) QTL analysis of leaf morphology indicates conserved shape loci in grapevine. Front Plant Sci. 10:1373. DOI: https://doi.org/10.3389/fpls.2019.01373
- Type:
Journal Articles
Status:
Published
Year Published:
2020
Citation:
Smith SY, Chitwood DH (2020) Plant-environment interactions: A sweeping perspective. Int J Plant Sci. 181(2). DOI: https://doi.org/10.1086/707481
- Type:
Journal Articles
Status:
Published
Year Published:
2020
Citation:
Rowland SD, Zumstein K, Nakayama H, Cheng Z, Flores AM, Chitwood DH, Maloof JN, Sinha NR (2020) Leaf shape is a predictor of fruit quality and cultivar performance in tomato. New Phytol. 226:851-865. DOI: https://doi.org/10.1111/nph.16403. Pre-print: https://doi.org/10.1101/584466
- Type:
Journal Articles
Status:
Published
Year Published:
2020
Citation:
Baumgartner A, Donahoo M, Chitwood DH, Peppe DJ (2020) The influences of environmental change and development on leaf shape in Vitis. American Journal of Botany. 107(4):1-13. DOI: https://doi.org/10.1002/ajb2.1460
- Type:
Journal Articles
Status:
Published
Year Published:
2019
Citation:
Eithun E, Chitwood DH, Larson J, Lang G, Munch E (2019) Isolating phyllotactic patterns embedded in the secondary growth of sweet cherry (Prunus avium L.) using magnetic resonance imaging. Plant Methods. 15, 111. DOI: https://doi.org/10.1186/s13007-019-0496-7. Pre-print: https://arxiv.org/abs/1812.03321
- Type:
Journal Articles
Status:
Published
Year Published:
2020
Citation:
Am�zquita EJ, Quigley MY, Ophelders T, Munch E, Chitwood DH (2020) The shape of things to come: Topological data analysis and biology, from molecules to organisms. Developmental Dynamics. 1-18. [Featured cover article, July issue]. DOI: https://doi.org/10.1002/dvdy.175
|
Progress 07/01/19 to 09/30/19
Outputs
Target Audience:Research-wise, target audiences have included both the plant science and applied mathematics communities. The proposed research seeks to measure plant morphology in exquisite detail using X-ray CT, so that this can be paired with genetic, environmental response, functional, and management data to fully understand the factors that affect the plant form. The above requires not only working closely with plant scientists but communicating their needs to applied mathematicians so that an underlying theory can be developed. Plant scientists and applied mathematician audiences have both been engaged during the reporting period, through collaboration, publications, and attending meetings. Students are another target audience that have been reached through both formal classroom instruction and the development of innovative teaching methodologies. Educating plant science students that are fluent in coding and mathematics is necessary to carry out the proposed research, but finding students of the required proficiency is difficult. During the reporting period, a 3 credit graduate level course was developed that teaches Python and mathematical modeling of plant morphology to students who do not necessarily have experience in either. The course is designed using a flipped classroom approach and is entirely coded in Python notebooks. The course materials have been made public. Over four hours of instructional videos were made as well. Engaging students and educating them to be proficient in Python coding and mathematical modeling is an indispensable part of the proposed research necessary to enable its goals. Changes/Problems:The biggest change was realizing that generating large, X-ray CT datasets was not as difficult as imagined. Generating X-ray CT data takes time, but it is routine and predictable. This is positive in that we have generated lots of data successfully, but we will be changing our focus to anlysis. Personnel-wise, graduate student Mitchell Eithun has decided to pursue a non-science career. Further, post-doc Dr. Tim Ophelders is expected to leave the project by the end of next reporting period. We will be addressing these changes in personnel in the next reporting period by recruiting additional members to the lab. What opportunities for training and professional development has the project provided?Post-doc/X-ray CT manager Dr. Michelle Quigley (70/30 Chitwood/Munch): Dr. Quigley has become proficient in X-ray imaging and manages the X-ray CT machine. She has become versed in the safety and operating protocols for the machine. Post-doc Dr. Tim Ophelders (30/70 Chitwood/Munch): Dr. Ophelders has mentored graduates students and undergraduates in computation and mathematical analyses of data. He has attended professional meetings related to Topological Data Analysis (TDA). Graduate student Mitchell Eithun (50/50 Chitwood/Munch): Mitchell Eithun attended a plant phenomics conference to increase his knowledge about plant biology. He was instrumental in developing image analysis techniques and code to analyze data during the reporting period, especially for the citrus dataset. Mitchell has decided not to pursue science and has left graduate school at the end of the reporting period. Graduate student Erik Amézquita (50/50 Chitwood/Munch): Erik Amézquita has attended several mathematical conferences and taken classes in both mathematics and plant biology to improve his skills applying TDA to the analysis of plant morphology. Erik has been instrumental in developing image anlaysis and TDA techniques for the barley project. Undergraduate Joey Mullins (100 Chitwood): Joey is part of the MSU Professorial Assistants program. Joey has learned how to operate the X-ray CT machine and assists Dr. Michelle Quigley in scanning projects. How have the results been disseminated to communities of interest?By coupling X-ray CT with virtual reality (VR) headsets an excellent outreach tool is created where viewers can interact with plant images in 3D. Dr. Michelle Quigley has created a VR headset set-up that has been used at the MSU Fascination with Plants Day and MSU Science Festival. She has also taken the VR set-up to area schools in Lansing. Together with Dr. Bob VanBuren, Dr. Chitwood has co-created a new 3 credit course to teach Python to those with no prior coding experience. The course focuses on teaching mathematical modeling of plant morphology and uses examples related to morphometrics. The course will help train plant biologists versed in coding to tackle the Objectives outlined in this proposal. The course has been made public to disseminate these materials broadly. What do you plan to do during the next reporting period to accomplish the goals?For the next reporting period, we will focus more of our energy on analysis of data. Image processing, which is separate from TDA, in particular has proven challenging, because the code is not generalizable between projects. For future reporting periods we will be focusing less on generating X-ray CT data and more on analyzing the data we have already generated.
Impacts
What was accomplished under these goals?
In a world of big data, in which we can measure vast quantities of information quickly, the primary problem we encounter is how to make sense of what we have measured. We think of data in terms of human-generated information, but the natural world contains multitudes more data. The form of every organism is exquisitely encoded by genetic information and sculpted by environmental forces. Evolutionary forces are responsible for feeding the world's civilizations through domestication, and environmental responses will be the key to mitigating the effects of climate change on our crops. If we could not only record everything that a plant is, but also quantify that information, so that we could make predictions using genetic and environmental variables, we would be in a better place to rationally-design the plants and agriculture of the future. The proposed work uses a combination of X-ray Computed Tomography (CT) and a mathematical field known as Topological Data Analysis (TDA) to record and measure the plant form comprehensively. We use industrial X-ray CT (just like the medical technology) to create 3D reconstructions of plants--both internally and externally--at extremely high resolution. We use mathematics to quantify the complex morphologies of plants, considering the entirety of data, from the organ to tissue levels in multiple dimensions. The resulting information, a measurement of what a plant is, is then modeled with genetic and environmental information. Below, some applications of the above work relating to the proposed major goals are outlined: Objective 1: Analyze 3D objects using a variety of filter functions, and compare the outputs using benchmarks such as heritability and predictability. Experiments conducted: During the research period, we have started working with numerous plant science collaborators, including: 1) tracking the effects of selection on barley grain morphology in a long-term evolution experiment, 2) studying the evolution of citrus morphology during evolution and in hybridization events, 3) creating time-lapses of Arabidopsis and succulent growth and integrating these with gene expression datasets, 4) predicting shelling properties of walnuts from their shape, and 5) reconstructing the domestication history of sunflower using ancient preserved discs and DNA. Data collected: For the above experiments conducted, we have generated large, X-ray CT datasets. Summary Statistics and summary of results: These datasets number from hundreds to thousands of X-ray CT scans a piece, typically a few terabytes of data total for each project. Each scan is a morphological reconstruction of the specimen, typically to ~50-100 micron resolution in 3D. Key outcomes: The data generated above is a change in condition. We have captured intensive phenotypic data of the morphology of plant specimens. This data will subsequently be used to study the evolution of crop morphology, the development of plant organs, and the underpinnings of desirable traits. Objective 2: Create a framework to analyze dynamic 4D objects through time. Experiments conducted: A key objective of the proposed work is to not only study static plant shapes, but dynamic ones as well, especially plant development. We have taken X-ray CT time-series of growing plants. Data collected: High-resolution, 3D X-ray CT images of growing plants were imaged daily in replication. Summary Statistics and summary of results: Several succulents and two Arabidopsis accessions with contrasting leaf shapes and developmental stability were imaged as a time series. Key outcomes: The analysis of this data is in process. The data will enable a change in knowledge as a pipeline for image processing, that not only accurately isolates individual plants and their organs, but registers these objects across time for TDA, is developed. We are applying for internal funding at MSU to combine these developmental series of plant morphology with gene expression networks as well. Objective 3: Explore Topological Data Analysis (TDA) methods to visualize, interpret, and disseminate X-ray CT data and morphological analyses. Experiments conducted: Most of the first year of this proposal has been used to create large datasets of X-ray CT images for subsequent analysis. Two of those datasets--barley spikes from an artificial evolution experiment and scans of citrus from the USDA-UC Riverside germplasm--have been preliminarily analyzed using TDA. Data collected: Image analysis techniques were developed to allow the processing of X-ray CT images. Summary Statistics and summary of results: For both the barley and citrus datasets, techniques to accurately segment objects away from air and correctly label individual items in a single scan were developed so that each barley spike and citrus fruit is saved as an independent file. Segmentation algorithms were developed to isolate internal features within each barley spike and citrus fruit. For the barley spikes, the rachis, seed, seed coat, and endosperm were isolated. For the citrus, the exocarp, mesocarp, and endocarp of each fruit was isolated. For the barley seeds, a distance function, X-ray intensity, and geodesic distance filter were used to calculate Euler Characteristic Curves (ECCs). The ECCs are a quantification of shape using TDA. Key outcomes: Applying image processing techniques and TDA to two large datasets, we achieved a change in knowledge. The ability to handle and analyze large, voxel-based, 3D images is challenging, and we hope to expand our techniques and analysis not only for the barley and citrus datasets, but for the remainder of X-ray CT images we have collected during the next reporting period.
Publications
- Type:
Conference Papers and Presentations
Status:
Published
Year Published:
2019
Citation:
Chitwood DH, Eithun M, Munch E, Ophelders T. (2019) Topological Mapper for 3D Volumetric Images. In: Burgeth B., Kleefeld A., Naegel B., Passat N., Perret B. (eds) Mathematical Morphology and Its Applications to Signal and Image Processing. ISMM 2019. Lecture Notes in Computer Science, vol 11564. Springer, Cham. DOI: https://doi.org/10.1007/978-3-030-20867-7_7
- Type:
Journal Articles
Status:
Published
Year Published:
2019
Citation:
Migicovsky Z, Harris ZN, Klein LL, Li M, McDermaid A, Chitwood DH, Fennell A, Kovacs LG, Kwasniewski M, Londo JP, Ma Q, Miller AJ (2019) Rootstock effects on scion phenotypes in a Chambourcin experimental vineyard. Hort. Res. 6, 64. DOI: https://doi.org/10.1038/s41438-019-0146-2
|