Application of machine learning algorithms in analysing single cell/nuclei RNA sequencing datasets in rat model of Parkinson's disease
(2021) BINP52 20202Degree Projects in Bioinformatics
- Abstract
- Parkinson’s disease (PD) is the most common neurodegenerative disorder, characterized by slow progressive loss of dopamine (DA) neurons in the substantia nigra, located in the ventral midbrain. Although the relatively focal degeneration in PD makes it a good candidate for cell-based therapies, obtaining cell source on large scale clinical application remains a challenge. It has now been shown that functional dopamine neurons can be generated using stem cells and these cells are both safe and functional when transplanted in animal models of Parkinson’s disease. Intensive research efforts have been made in recent years to understand the molecular mechanisms controlling the developmental program and differentiation of DA neurons using... (More)
- Parkinson’s disease (PD) is the most common neurodegenerative disorder, characterized by slow progressive loss of dopamine (DA) neurons in the substantia nigra, located in the ventral midbrain. Although the relatively focal degeneration in PD makes it a good candidate for cell-based therapies, obtaining cell source on large scale clinical application remains a challenge. It has now been shown that functional dopamine neurons can be generated using stem cells and these cells are both safe and functional when transplanted in animal models of Parkinson’s disease. Intensive research efforts have been made in recent years to understand the molecular mechanisms controlling the developmental program and differentiation of DA neurons using advanced technologies like single cell RNA sequencing. Transcriptional profiling of these cells will provide a better cellular composition of stem cell based cell replacement therapy product. It will also help in determining the diversity of dopaminergic neurons, which is so far largely unknown due to the absence of molecular markers. Altogether these findings will help us understanding cell identity and molecular mechanisms underlying cell fate decisions of ventral midbrain neuron sub type during stem cell differentiation. However, fast paced growth in generation of these datasets also requires better optimization and development of bioinformatics algorithms. Here in this thesis, we focus on analyzing and integrating single cell and single nuclei datasets obtained from grafted cells sequenced after different time points of transplantation in a rat model of PD. We developed a machine learning based model to integrate these large-scale datasets, which can be further trained and easily distributed and replicated to other similar studies in the field (Less)
- Popular Abstract
- Application of machine learning algorithms and analysis of sc/sn RNA sequencing datasets in rat model of Parkinson’s disease
Parkinson’s disease (PD) is the most common neurodegenerative disorder, characterized by slow progressive loss of dopamine (DA) neurons in midbrain. Although the relatively focal degeneration in PD makes it a good candidate for cell-based therapies, obtaining cell source on large scale clinical application remains a challenge. Next generation sequencing technologies such as single cell RNA sequencing have allowed us to transcriptionally profile individual cells for later use as a therapy.
Given the fact that the affected neuron population is so homogeneous and localised in one area of the brain, cell... (More) - Application of machine learning algorithms and analysis of sc/sn RNA sequencing datasets in rat model of Parkinson’s disease
Parkinson’s disease (PD) is the most common neurodegenerative disorder, characterized by slow progressive loss of dopamine (DA) neurons in midbrain. Although the relatively focal degeneration in PD makes it a good candidate for cell-based therapies, obtaining cell source on large scale clinical application remains a challenge. Next generation sequencing technologies such as single cell RNA sequencing have allowed us to transcriptionally profile individual cells for later use as a therapy.
Given the fact that the affected neuron population is so homogeneous and localised in one area of the brain, cell replacement therapy can be very beneficial. Past clinical trial have shown that grafting fetal dopaminergic neurons into patients caudate-putamen nuclei improve motor function and are able to survive. To move to large-scale clinical applications, the current challenge is to recreate authentic and functional dopamine neurons from human embryonic stem cells (hESCs) in vitro, thereby opening up unprecedented opportunities to gain access to a renewable source of cells potentially suitable for PD therapeutic applications.
A crucial component of these scientific advances has been the parallel development of bioinformatics and sequencing applications that allow us to analyse cell cultures and determine the different cell types that are present, graft or host population. But these advances are bringing up a new set of problems to solve such as lack of integrability between datasets, due to sequencing, technological, conditional or species derived batch effects. One way to solve this could be by manually curating and analysing data, which brings another issue, the considerable amount of human resources, less reproducibility and time needed to perform these tasks.
Recent developments in the field of machine learning and data science could be very helpful in addressing these limitations. Using machine learning we can create models that allow us to transfer knowledge in the shape of labels containing the desired information from a reference dataset to a new one and if trained properly, integrating datasets across platforms, reducing the amount of time required to analyse a whole dataset from days to just hours on a powerful enough machine.
We created an autoencoder based model that allows us to transfer cell type labels from the grafted host cell to future single cell or single nuclei datasets, as well as integrating both technologies together.
Master’s Degree Bioinformatics 60 credits 2021
Department of Biology, Lund University
Advisor: Yogita Sharma
Advisors Developmental and regenerative neurobiology (Less)
Please use this url to cite or link to this publication:
http://lup.lub.lu.se/student-papers/record/9060852
- author
- Hildago Gil, David
- supervisor
- organization
- course
- BINP52 20202
- year
- 2021
- type
- H2 - Master's Degree (Two Years)
- subject
- language
- English
- id
- 9060852
- date added to LUP
- 2021-07-05 10:37:55
- date last changed
- 2021-07-05 10:37:55
@misc{9060852,
abstract = {{Parkinson’s disease (PD) is the most common neurodegenerative disorder, characterized by slow progressive loss of dopamine (DA) neurons in the substantia nigra, located in the ventral midbrain. Although the relatively focal degeneration in PD makes it a good candidate for cell-based therapies, obtaining cell source on large scale clinical application remains a challenge. It has now been shown that functional dopamine neurons can be generated using stem cells and these cells are both safe and functional when transplanted in animal models of Parkinson’s disease. Intensive research efforts have been made in recent years to understand the molecular mechanisms controlling the developmental program and differentiation of DA neurons using advanced technologies like single cell RNA sequencing. Transcriptional profiling of these cells will provide a better cellular composition of stem cell based cell replacement therapy product. It will also help in determining the diversity of dopaminergic neurons, which is so far largely unknown due to the absence of molecular markers. Altogether these findings will help us understanding cell identity and molecular mechanisms underlying cell fate decisions of ventral midbrain neuron sub type during stem cell differentiation. However, fast paced growth in generation of these datasets also requires better optimization and development of bioinformatics algorithms. Here in this thesis, we focus on analyzing and integrating single cell and single nuclei datasets obtained from grafted cells sequenced after different time points of transplantation in a rat model of PD. We developed a machine learning based model to integrate these large-scale datasets, which can be further trained and easily distributed and replicated to other similar studies in the field}},
author = {{Hildago Gil, David}},
language = {{eng}},
note = {{Student Paper}},
title = {{Application of machine learning algorithms in analysing single cell/nuclei RNA sequencing datasets in rat model of Parkinson's disease}},
year = {{2021}},
}