{"id":{"repo_id":"cambridge","oai_identifier":"oai:www.repository.cam.ac.uk:1810/391305"},"canonical_url":"https://search.dev.ndltd.org/etd/cambridge/oai:www.repository.cam.ac.uk:1810/391305","repository":{"repo_id":"cambridge","name":"Cambridge University","base_url":"https://api.repository.cam.ac.uk/server/oai/request"},"display":{"title":"A Neural Network and Genome Language Model Based Approach to Understanding DNA Replication in Cancer","abstract":"The rapid acceleration of technological innovation over the last two decades has catalysed an unprecedented surge in biological data generation, exemplified by the sequencing of the human genome with its ~6.4 billion base pairs, enough to fill nearly two million pages if printed out for a single person. One fundamental process of life is copying the human genome during cell division to give rise to two genetically identical daughter cells. This process is called DNA replication. DNA replication differs between healthy cells and cancer cells and can be exploited as a therapeutic target. At the same time, altered replication dynamics are critical for understanding cancer onset, evolution and treatment resistance. DNA replication during cell division is a tightly regulated process, and maintaining genome integrity requires complete, error-free duplication of the DNA. In this PhD thesis, I leveraged our development of an artificial intelligence-based method called DNAscent, which enables high-throughput, single-molecule analysis of the DNA replication dynamics at high spatial resolution to understand aberrant DNA replication in different cancer models. To improve our understanding of the initiation of DNA replication, I developed a genome language model to determine from which regions of the genome DNA replication can start. This PhD thesis is divided into three major projects: Chapter 2: Replication of extrachromosomal DNA Recently, the role of extrachromosomal DNA (ecDNA) in cancer has gained a lot of attention because ecDNA often harbours oncogene amplifications, driving rapid tumour evolution, and occurs in ~17 % of all newly diagnosed cancers. I used DNAscent to study the DNA replication dynamics on ecDNA and identify differences between ecDNA and chromosomal DNA. I compared replication fork velocities and stall scores on single molecules between an ecDNA-positive and an ecDNA-negative colorectal cancer cell line derived from the same tumour. My findings indicate that DNA replication dynamics differ between ecDNA and chromosomal DNA and the presence of ecDNA affects the replication dynamics on chromosomal DNA. This study further suggests that ecDNA replication could be disrupted by therapeutic agents, offering a way to target cancer cells with oncogene amplifications on ecDNA. Chapter 3: DNA replication dynamics under ATR inhibitor treatment in a panel of six breast cancer cell lines Multiple clinical trials are underway to assess the performance of novel chemotherapeutics targeting the DNA replication stress response pathway. A central regulator of the replication stress response is the protein kinase ATR. Using our software DNAscent, I investigated the effect of ATR inhibition (ceralasertib) on the DNA replication dynamics in a panel of six breast cancer cell lines. Furthermore, I explored whether the DNA replication dynamics in the untreated group could be used to predict the ATR inhibitor response and find that they have predictive power. The results indicate that there are differences in DNA replication dynamics between the six breast cancer cell lines preceding the treatment with ATR inhibitor, which could be exploited to predict ATR inhibitor treatment outcome. Chapter 4: A novel genome language model to detect human origin of replication sequences Origin firing is a central process during DNA replication, but a sequence dependence in human cells has yet to be elucidated. I leveraged recent developments in genome language models to train a model that determines which sequences in the human genome can act as origins of replication. My fine-tuned genome language model achieves high performance metrics when classifying input sequences as origins of replication. The work in this Chapter highlights the potential of genome language models for studying DNA replication in human cells and enables future research in understanding the interplay of DNA replication stress, aberrant origin firing and treatment response for replication stress response-targeting therapeutics. Overall, in my thesis I demonstrated the importance of DNA replication dynamics features for understanding and treating cancer. The findings in my thesis pave the way towards establishing DNA replication targeting therapeutics for ecDNA-positive cancers, developing more comprehensive biomarkers for treatment response prediction for ATR inhibitors to capture the complex underlying biology and show a way towards integrating insights from genome language models to inform future research to elucidate where replication can be initiated in the human genome.","abstract_html":"The rapid acceleration of technological innovation over the last two decades has catalysed an unprecedented surge in biological data generation, exemplified by the sequencing of the human genome with its ~6.4 billion base pairs, enough to fill nearly two million pages if printed out for a single person. One fundamental process of life is copying the human genome during cell division to give rise to two genetically identical daughter cells. This process is called DNA replication. DNA replication differs between healthy cells and cancer cells and can be exploited as a therapeutic target. At the same time, altered replication dynamics are critical for understanding cancer onset, evolution and treatment resistance. DNA replication during cell division is a tightly regulated process, and maintaining genome integrity requires complete, error-free duplication of the DNA. In this PhD thesis, I leveraged our development of an artificial intelligence-based method called DNAscent, which enables high-throughput, single-molecule analysis of the DNA replication dynamics at high spatial resolution to understand aberrant DNA replication in different cancer models. To improve our understanding of the initiation of DNA replication, I developed a genome language model to determine from which regions of the genome DNA replication can start. This PhD thesis is divided into three major projects: Chapter 2: Replication of extrachromosomal DNA Recently, the role of extrachromosomal DNA (ecDNA) in cancer has gained a lot of attention because ecDNA often harbours oncogene amplifications, driving rapid tumour evolution, and occurs in ~17 % of all newly diagnosed cancers. I used DNAscent to study the DNA replication dynamics on ecDNA and identify differences between ecDNA and chromosomal DNA. I compared replication fork velocities and stall scores on single molecules between an ecDNA-positive and an ecDNA-negative colorectal cancer cell line derived from the same tumour. My findings indicate that DNA replication dynamics differ between ecDNA and chromosomal DNA and the presence of ecDNA affects the replication dynamics on chromosomal DNA. This study further suggests that ecDNA replication could be disrupted by therapeutic agents, offering a way to target cancer cells with oncogene amplifications on ecDNA. Chapter 3: DNA replication dynamics under ATR inhibitor treatment in a panel of six breast cancer cell lines Multiple clinical trials are underway to assess the performance of novel chemotherapeutics targeting the DNA replication stress response pathway. A central regulator of the replication stress response is the protein kinase ATR. Using our software DNAscent, I investigated the effect of ATR inhibition (ceralasertib) on the DNA replication dynamics in a panel of six breast cancer cell lines. Furthermore, I explored whether the DNA replication dynamics in the untreated group could be used to predict the ATR inhibitor response and find that they have predictive power. The results indicate that there are differences in DNA replication dynamics between the six breast cancer cell lines preceding the treatment with ATR inhibitor, which could be exploited to predict ATR inhibitor treatment outcome. Chapter 4: A novel genome language model to detect human origin of replication sequences Origin firing is a central process during DNA replication, but a sequence dependence in human cells has yet to be elucidated. I leveraged recent developments in genome language models to train a model that determines which sequences in the human genome can act as origins of replication. My fine-tuned genome language model achieves high performance metrics when classifying input sequences as origins of replication. The work in this Chapter highlights the potential of genome language models for studying DNA replication in human cells and enables future research in understanding the interplay of DNA replication stress, aberrant origin firing and treatment response for replication stress response-targeting therapeutics. Overall, in my thesis I demonstrated the importance of DNA replication dynamics features for understanding and treating cancer. The findings in my thesis pave the way towards establishing DNA replication targeting therapeutics for ecDNA-positive cancers, developing more comprehensive biomarkers for treatment response prediction for ATR inhibitors to capture the complex underlying biology and show a way towards integrating insights from genome language models to inform future research to elucidate where replication can be initiated in the human genome.","abstract_has_math":false,"creators":["Pfuderer, Pauline Luise"],"institution":"University of Cambridge","degree_name":"Doctor of Philosophy (PhD)","degree_level":"Doctoral","degree_discipline":null,"degree_department":null,"school":null,"contributors":[],"advisors":["Boemo, Michael"],"committee_chairs":[],"committee_members":[],"year":2025,"date_issued":"2025-09-02","date_published":"2025-09-02","updated_at":"2026-07-24T01:33:23Z","subjects":["Artificial Intelligence","Cancer"],"languages":["eng"],"rights":[],"rights_urls":["https://www.repository.cam.ac.uk/bitstreams/83cac4f1-b8b6-4eb9-b334-ee1fd8997324/download","http://purl.org/NET/rdflicense/allrightsreserved"],"identifier_entries":[]},"links":{"outbound_url":"https://doi.org/10.17863/CAM.122482","outbound_label":"DOI","outbound_source":"dc:identifier.doi"},"metadata_groups":[{"id":"people","label":"People","entries":[{"key":"dc:contributor.advisor","label":"Advisor","values":["Boemo, Michael"]},{"key":"dc:contributor.sponsor","label":"Sponsor","values":["CRUK Cambridge Centre PhD Studentship, Cambridge Trust International Scholarship (Honorary), Peterhouse Muir Wood Fund Award, The Alan Turing Institute PhD Enrichment Scheme"]},{"key":"dc:creator","label":"Author","values":["Pfuderer, Pauline Luise"]}]},{"id":"academic_context","label":"Academic Context","entries":[{"key":"dc:date.issued","label":"Date","values":["2025-09-02"]},{"key":"dc:publisher.institution","label":"Dc Publisher Institution","values":["University of Cambridge"]},{"key":"dc:relation.isreferencedby.uri","label":"Dc Relation Isreferencedby URI","values":["https://www.repository.cam.ac.uk/handle/1810/391305"]},{"key":"dc:type","label":"Dc Type","values":["Thesis"]},{"key":"dc:type.qualificationlevel","label":"Dc Type Qualificationlevel","values":["Doctoral"]},{"key":"dc:type.qualificationname","label":"Dc Type Qualificationname","values":["Doctor of Philosophy (PhD)"]}]},{"id":"subjects_keywords","label":"Subjects and Keywords","entries":[{"key":"dc:subject","label":"Dc Subject","values":["Artificial Intelligence","Cancer"]}]},{"id":"language_rights","label":"Language and Rights","entries":[{"key":"dc:language","label":"Dc Language","values":["eng"]},{"key":"dc:rights","label":"Dc Rights","values":["https://www.repository.cam.ac.uk/bitstreams/83cac4f1-b8b6-4eb9-b334-ee1fd8997324/download","http://purl.org/NET/rdflicense/allrightsreserved"]},{"key":"dc:rights.embargodate","label":"Dc Rights Embargodate","values":["2026-10-22"]},{"key":"dc:rights.embargotype","label":"Dc Rights Embargotype","values":["embargo"]}]},{"id":"identifiers","label":"Identifiers","entries":[{"key":"dc:identifier.doi","label":"DOI","values":["https://doi.org/10.17863/CAM.122482"]},{"key":"dc:identifier.uri","label":"Identifier URI","values":["https://www.repository.cam.ac.uk/bitstreams/eaf6807b-3bf8-48f6-b662-188cc710adac/download"]}]},{"id":"additional","label":"Additional Metadata","entries":[{"key":"dc:description.abstract","label":"Abstract","values":["The rapid acceleration of technological innovation over the last two decades has catalysed an unprecedented surge in biological data generation, exemplified by the sequencing of the human genome with its ~6.4 billion base pairs, enough to fill nearly two million pages if printed out for a single person. One fundamental process of life is copying the human genome during cell division to give rise to two genetically identical daughter cells. This process is called DNA replication. DNA replication differs between healthy cells and cancer cells and can be exploited as a therapeutic target. At the same time, altered replication dynamics are critical for understanding cancer onset, evolution and treatment resistance. DNA replication during cell division is a tightly regulated process, and maintaining genome integrity requires complete, error-free duplication of the DNA. In this PhD thesis, I leveraged our development of an artificial intelligence-based method called DNAscent, which enables high-throughput, single-molecule analysis of the DNA replication dynamics at high spatial resolution to understand aberrant DNA replication in different cancer models. To improve our understanding of the initiation of DNA replication, I developed a genome language model to determine from which regions of the genome DNA replication can start. This PhD thesis is divided into three major projects: Chapter 2: Replication of extrachromosomal DNA Recently, the role of extrachromosomal DNA (ecDNA) in cancer has gained a lot of attention because ecDNA often harbours oncogene amplifications, driving rapid tumour evolution, and occurs in ~17 % of all newly diagnosed cancers. I used DNAscent to study the DNA replication dynamics on ecDNA and identify differences between ecDNA and chromosomal DNA. I compared replication fork velocities and stall scores on single molecules between an ecDNA-positive and an ecDNA-negative colorectal cancer cell line derived from the same tumour. My findings indicate that DNA replication dynamics differ between ecDNA and chromosomal DNA and the presence of ecDNA affects the replication dynamics on chromosomal DNA. This study further suggests that ecDNA replication could be disrupted by therapeutic agents, offering a way to target cancer cells with oncogene amplifications on ecDNA. Chapter 3: DNA replication dynamics under ATR inhibitor treatment in a panel of six breast cancer cell lines Multiple clinical trials are underway to assess the performance of novel chemotherapeutics targeting the DNA replication stress response pathway. A central regulator of the replication stress response is the protein kinase ATR. Using our software DNAscent, I investigated the effect of ATR inhibition (ceralasertib) on the DNA replication dynamics in a panel of six breast cancer cell lines. Furthermore, I explored whether the DNA replication dynamics in the untreated group could be used to predict the ATR inhibitor response and find that they have predictive power. The results indicate that there are differences in DNA replication dynamics between the six breast cancer cell lines preceding the treatment with ATR inhibitor, which could be exploited to predict ATR inhibitor treatment outcome. Chapter 4: A novel genome language model to detect human origin of replication sequences Origin firing is a central process during DNA replication, but a sequence dependence in human cells has yet to be elucidated. I leveraged recent developments in genome language models to train a model that determines which sequences in the human genome can act as origins of replication. My fine-tuned genome language model achieves high performance metrics when classifying input sequences as origins of replication. The work in this Chapter highlights the potential of genome language models for studying DNA replication in human cells and enables future research in understanding the interplay of DNA replication stress, aberrant origin firing and treatment response for replication stress response-targeting therapeutics. Overall, in my thesis I demonstrated the importance of DNA replication dynamics features for understanding and treating cancer. The findings in my thesis pave the way towards establishing DNA replication targeting therapeutics for ecDNA-positive cancers, developing more comprehensive biomarkers for treatment response prediction for ATR inhibitors to capture the complex underlying biology and show a way towards integrating insights from genome language models to inform future research to elucidate where replication can be initiated in the human genome."]},{"key":"dc:format.checksum.md5","label":"Dc Format Checksum Md5","values":["65992508924774f4440abdf30ad50b2b","87eda9de84448d1f82354d60eee3eb5f"]},{"key":"dc:title","label":"Title","values":["A Neural Network and Genome Language Model Based Approach to Understanding DNA Replication in Cancer"]}]}],"canonical_facts":{"dc:contributor.advisor":["Boemo, Michael"],"dc:contributor.sponsor":["CRUK Cambridge Centre PhD Studentship, Cambridge Trust International Scholarship (Honorary), Peterhouse Muir Wood Fund Award, The Alan Turing Institute PhD Enrichment Scheme"],"dc:creator":["Pfuderer, Pauline Luise"],"dc:date.issued":["2025-09-02"],"dc:description.abstract":["The rapid acceleration of technological innovation over the last two decades has catalysed an unprecedented surge in biological data generation, exemplified by the sequencing of the human genome with its ~6.4 billion base pairs, enough to fill nearly two million pages if printed out for a single person. One fundamental process of life is copying the human genome during cell division to give rise to two genetically identical daughter cells. This process is called DNA replication. DNA replication differs between healthy cells and cancer cells and can be exploited as a therapeutic target. At the same time, altered replication dynamics are critical for understanding cancer onset, evolution and treatment resistance. DNA replication during cell division is a tightly regulated process, and maintaining genome integrity requires complete, error-free duplication of the DNA. In this PhD thesis, I leveraged our development of an artificial intelligence-based method called DNAscent, which enables high-throughput, single-molecule analysis of the DNA replication dynamics at high spatial resolution to understand aberrant DNA replication in different cancer models. To improve our understanding of the initiation of DNA replication, I developed a genome language model to determine from which regions of the genome DNA replication can start. This PhD thesis is divided into three major projects: Chapter 2: Replication of extrachromosomal DNA Recently, the role of extrachromosomal DNA (ecDNA) in cancer has gained a lot of attention because ecDNA often harbours oncogene amplifications, driving rapid tumour evolution, and occurs in ~17 % of all newly diagnosed cancers. I used DNAscent to study the DNA replication dynamics on ecDNA and identify differences between ecDNA and chromosomal DNA. I compared replication fork velocities and stall scores on single molecules between an ecDNA-positive and an ecDNA-negative colorectal cancer cell line derived from the same tumour. My findings indicate that DNA replication dynamics differ between ecDNA and chromosomal DNA and the presence of ecDNA affects the replication dynamics on chromosomal DNA. This study further suggests that ecDNA replication could be disrupted by therapeutic agents, offering a way to target cancer cells with oncogene amplifications on ecDNA. Chapter 3: DNA replication dynamics under ATR inhibitor treatment in a panel of six breast cancer cell lines Multiple clinical trials are underway to assess the performance of novel chemotherapeutics targeting the DNA replication stress response pathway. A central regulator of the replication stress response is the protein kinase ATR. Using our software DNAscent, I investigated the effect of ATR inhibition (ceralasertib) on the DNA replication dynamics in a panel of six breast cancer cell lines. Furthermore, I explored whether the DNA replication dynamics in the untreated group could be used to predict the ATR inhibitor response and find that they have predictive power. The results indicate that there are differences in DNA replication dynamics between the six breast cancer cell lines preceding the treatment with ATR inhibitor, which could be exploited to predict ATR inhibitor treatment outcome. Chapter 4: A novel genome language model to detect human origin of replication sequences Origin firing is a central process during DNA replication, but a sequence dependence in human cells has yet to be elucidated. I leveraged recent developments in genome language models to train a model that determines which sequences in the human genome can act as origins of replication. My fine-tuned genome language model achieves high performance metrics when classifying input sequences as origins of replication. The work in this Chapter highlights the potential of genome language models for studying DNA replication in human cells and enables future research in understanding the interplay of DNA replication stress, aberrant origin firing and treatment response for replication stress response-targeting therapeutics. Overall, in my thesis I demonstrated the importance of DNA replication dynamics features for understanding and treating cancer. The findings in my thesis pave the way towards establishing DNA replication targeting therapeutics for ecDNA-positive cancers, developing more comprehensive biomarkers for treatment response prediction for ATR inhibitors to capture the complex underlying biology and show a way towards integrating insights from genome language models to inform future research to elucidate where replication can be initiated in the human genome."],"dc:format.checksum.md5":["65992508924774f4440abdf30ad50b2b","87eda9de84448d1f82354d60eee3eb5f"],"dc:identifier.doi":["https://doi.org/10.17863/CAM.122482"],"dc:identifier.uri":["https://www.repository.cam.ac.uk/bitstreams/eaf6807b-3bf8-48f6-b662-188cc710adac/download"],"dc:language":["eng"],"dc:publisher.institution":["University of Cambridge"],"dc:relation.isreferencedby.uri":["https://www.repository.cam.ac.uk/handle/1810/391305"],"dc:rights":["https://www.repository.cam.ac.uk/bitstreams/83cac4f1-b8b6-4eb9-b334-ee1fd8997324/download","http://purl.org/NET/rdflicense/allrightsreserved"],"dc:rights.embargodate":["2026-10-22"],"dc:rights.embargotype":["embargo"],"dc:subject":["Artificial Intelligence","Cancer"],"dc:title":["A Neural Network and Genome Language Model Based Approach to Understanding DNA Replication in Cancer"],"dc:type":["Thesis"],"dc:type.qualificationlevel":["Doctoral"],"dc:type.qualificationname":["Doctor of Philosophy (PhD)"]},"updated_at":"2026-07-24T01:33:23Z"}