國立高雄大學圖資館 |

Language: English

Back

Discovering Biophysical Principles in Latent Space Representations of Immune Recognition.

Record Type:	Electronic resources : Monograph/item
Title/Author:	Discovering Biophysical Principles in Latent Space Representations of Immune Recognition.
Author:	Rademaker, Thomas J. .
Published:	Ann Arbor : ProQuest Dissertations & Theses, 2021
Description:	207 p.
Notes:	Source: Dissertations Abstracts International, Volume: 83-03, Section: B.
Notes:	Advisor: Francois, Paul.
Contained By:	Dissertations Abstracts International83-03B.
Subject:	Pathogens.
Online resource:	http://pqdd.sinica.edu.tw/twdaoapp/servlet/advanced?query=28731037
ISBN:	9798544222897

Discovering Biophysical Principles in Latent Space Representations of Immune Recognition.
Rademaker, Thomas J. .

Discovering Biophysical Principles in Latent Space Representations of Immune Recognition. - Ann Arbor : ProQuest Dissertations & Theses, 2021 - 207 p.

Source: Dissertations Abstracts International, Volume: 83-03, Section: B.

Thesis (Ph.D.)--McGill University (Canada), 2021.

This item must not be sold to any third party vendors.

The adaptive immune system is a complex biological system acting over many length and time scales. T cells, effector cells of the adaptive immune system, are capable of recognizing minute amounts of pathogens and mounting a grotesque response, while not responding at all to large amounts of self antigens. The response is tightly regulated at the intra-, inter- and extracellular level through intricate protein-protein interaction networks. While the interactions have been described qualitatively, a quantitative understanding is often lacking.In this thesis, we present computational approaches inspired by physics and ma- chine learning to quantitatively study different aspects of immune recognition. First, using fitness-based parameter reduction, we extract the core module from the intracellular network of immune recognition. Second, using machine learning techniques, we study the sensitivity of immune recognition networks to antagonism, a perturbation to the antigen distribution that prevents T cells from responding to pathogens. We find that the output function of robust immune recognition networks contains a critical point, a finding that informs the design of robust machine learning classifiers.Finally, we predict antigen quality from cytokine dynamics. We represent the cytokine profile in a latent space and parameterize the latent space using piecewise ballistic mod- els. We validate our model against diverse experimental configurations, providing us with a biological basis for the model parameters. Using these parameters, we predict antigen quality independent of antigen quantity and initial T cell number, providing a reference antigen quality that known baselines cannot provide with a single measurement.

ISBN: 9798544222897Subjects--Topical Terms:

915822
Pathogens.

Discovering Biophysical Principles in Latent Space Representations of Immune Recognition.
LDR:04996nmm a2200349 4500 001 616477
005 20220513114347.5
008 220920s2021 ||||||||||||||||| ||eng d
020 $a 9798544222897
035 $a (MiAaPQ)AAI28731037
035 $a (MiAaPQ)McGill_47429g25m
035 $a AAI28731037
040 $a MiAaPQ $c MiAaPQ
100 1 $a Rademaker, Thomas J. . $3 915821
245 1 0 $a Discovering Biophysical Principles in Latent Space Representations of Immune Recognition.
260 1 $a Ann Arbor : $b ProQuest Dissertations & Theses, $c 2021
300 $a 207 p.
500 $a Source: Dissertations Abstracts International, Volume: 83-03, Section: B.
500 $a Advisor: Francois, Paul.
502 $a Thesis (Ph.D.)--McGill University (Canada), 2021.
506 $a This item must not be sold to any third party vendors.
520 $a The adaptive immune system is a complex biological system acting over many length and time scales. T cells, effector cells of the adaptive immune system, are capable of recognizing minute amounts of pathogens and mounting a grotesque response, while not responding at all to large amounts of self antigens. The response is tightly regulated at the intra-, inter- and extracellular level through intricate protein-protein interaction networks. While the interactions have been described qualitatively, a quantitative understanding is often lacking.In this thesis, we present computational approaches inspired by physics and ma- chine learning to quantitatively study different aspects of immune recognition. First, using fitness-based parameter reduction, we extract the core module from the intracellular network of immune recognition. Second, using machine learning techniques, we study the sensitivity of immune recognition networks to antagonism, a perturbation to the antigen distribution that prevents T cells from responding to pathogens. We find that the output function of robust immune recognition networks contains a critical point, a finding that informs the design of robust machine learning classifiers.Finally, we predict antigen quality from cytokine dynamics. We represent the cytokine profile in a latent space and parameterize the latent space using piecewise ballistic mod- els. We validate our model against diverse experimental configurations, providing us with a biological basis for the model parameters. Using these parameters, we predict antigen quality independent of antigen quantity and initial T cell number, providing a reference antigen quality that known baselines cannot provide with a single measurement.
520 $a Le systeme immunitaire adaptatif est un systeme biologique complexe fonctionnant sur de nombreuses longueurs et echelles de temps. Les cellules T, cellules effectrices du systeme immunitaire adaptatif, sont capables de reconnaitre des quantites infimes d'agents pathogenes et de monter une reponse tres fort, tout en ne repondant pas du tout a de grandes quantites de soi-antigenes. La reponse est etroitement regulee au niveau intra-, inter- et extracellulaire par des reseaux complexes d'interaction proteine-proteine. Bien que les interactions aient ete decrites de maniere qualitative, une comprehension quantitative fait souvent defaut. Dans cette these, nous presentons des approches informatiques inspirees de la physique et de l'apprentissage automatique pour etudier quantitativement differents aspects de la reconnaissance immunitaire. Premierement, en utilisant la reduction des parametres basee sur la fitness, nous extrayons le module de base du reseau intracellulaire de reconnaissance immunitaire. Deuxiemement, a l'aide de techniques d'apprentissage automatique, nous etudions la sensibilite des reseaux de reconnaissance immunitaire a l'antagonisme, une perturbation de la distribution de l'antigene qui empeche les cellules T de repondre aux agents pathogenes. Nous constatons que la fonction de sortie des reseaux de reconnaissance immunitaire robustes contient un point critique, une decouverte qui informe la conception de classificateurs d'apprentissage automatique robustes. Enfin, nous predisons la qualite de l'antigene a partir des dynamiques de cytokines, molecules messageres extracellulaires. Nous representons le profil des cytokines dans un espace latent, parametrons l'espace latent a l'aide de modeles balistiques par morceaux et etudions des configurations experimentales, a partir desquelles nous extrayons une base biologique pour les parametres du modele. A partir de ces parametres, nous predisons la qualite de l'antigene independamment de la quantite d'antigene et du nombre initial de lymphocytes T, fournissant une qualite d'antigene de reference que les lignes de base connues ne peuvent pas fournir avec une seule mesure.
590 $a School code: 0781.
650 4 $a Pathogens. $3 915822
650 4 $a Biologists. $3 915823
650 4 $a Microprocessors. $3 182447
650 4 $a Cytokines. $3 195178
650 4 $a Immunology. $3 189152
650 4 $a Adaptation. $3 915824
650 4 $a Neurosciences. $3 211508
650 4 $a Biology. $3 188667
650 4 $a Peptides. $3 202659
650 4 $a Defense. $3 915825
650 4 $a Feedback. $3 915826
650 4 $a Editing. $3 179494
650 4 $a Proteins. $3 193852
650 4 $a T cell receptors. $3 915827
650 4 $a Signal transduction. $3 915634
650 4 $a Gene expression. $3 188888
650 4 $a Dendritic cells. $3 195318
650 4 $a Immune system. $3 195094
650 4 $a Neural networks. $3 915594
650 4 $a Medical research. $2 bicssc $3 533145
650 4 $a Lymphocytes. $3 193734
650 4 $a Ligands. $3 262924
650 4 $a Antigens. $3 246106
650 4 $a Algorithms. $3 184661
650 4 $a Genetics. $3 189015
650 4 $a Interdisciplinary aspects. $3 915828
690 $a 0982
690 $a 0306
690 $a 0317
690 $a 0369
710 2 $a McGill University (Canada). $3 531188
773 0 $t Dissertations Abstracts International $g 83-03B.
790 $a 0781
791 $a Ph.D.
792 $a 2021
793 $a English
856 4 0 $u http://pqdd.sinica.edu.tw/twdaoapp/servlet/advanced?query=28731037