Tumor Genetics Blog

Suomennos: Syöpäsolujen sosiaalisten verkostojen selvittelyä – mikä solujen sisäisessä kommunikaatiossa menee pieleen?

By Sini Hakkola

What doesn’t kill you, gives you a bunch of unhealthy survival mechanisms

Networks formed by people are complex and occasionally, communication within the network can go awry. Imagine a group of friends or a social media community, where arguments and misunderstanding arise from interactions. Sometimes it can be challenging to determine, which factors contributed to the conflict and why the discussion started to distort. This metaphor leads us to the topic of discussing conflicts in communication within gene regulatory networks in cells and how they can be studied using computational methods. Understanding these conflicts is important, as they are known to enable cells to divide uncontrollably, leading to cancer.

Gene regulatory networks are complex systems of genes, transcription factors and regulatory elements in DNA. The communication within these networks is altered in cancer, leading to abnormal behavior of the cells, such as uncontrolled growth, evasion of death signals, and resistance to treatments. In normal, healthy cells gene regulatory networks function in harmony to maintain cellular functions and homeostasis. However, in cancer, aberrations in gene regulatory networks lead to dysfunctional gene expression patterns that promote the development and progression of cancer.

Feeling upset about their own personal problems, one might snap at a customer service representative in a store, who in turn vents their work stress by snapping at their grandmother on the phone, who proceeds to lecture her upstairs neighbor for noisy stomping. This does not just spread distress to the immediate target of their anger but also propagates it further within a complex network of interactions. In a somewhat similar manner in cells, the emergence of aberrations triggers the progression of its effects in a chain reaction-like manner throughout the gene regulatory network and results in the development of pathological survival mechanisms. The initial aberrations can be, for example, gene mutations, amplifications or deletions or epigenetic alterations that do not affect the DNA sequence, but still affect the transcription of genes, such as changes in the openness of the DNA for the transcription machinery. Characterizing and understanding these aberrations is needed so that we can develop better cancer treatments in the future.

Cracking the sequence one cell at a time

Inferring gene regulatory networks in cancer is a complex task. This is due to high heterogeneity of cancer cells, meaning that different cells within the same tumor can possess different alterations and therefore, distinct abilities to promote the disease progression and, for example, resist treatments. Therefore, in the Computational Biology group, we are analyzing gene regulatory networks using single-cell sequencing data, which enables the high-resolution characterization of gene regulation on the level of single cancer cells. Specifically, we are interested in gene regulatory networks in the context of prostate cancer treatment response.

Single-cell transcriptome (RNA) sequencing data gives a snapshot of which genes are expressed within each cell at a specific moment. Complementarily, single-cell transposase-accessible chromatin (ATAC) sequencing measures accessibility of the chromatin, providing insights into which regions of DNA are accessible for transcription factors and other regulatory proteins to control gene expression. By analyzing the patterns of accessible chromatin in single cells, we can infer regulatory regions and understand their impact on gene expression. By integrating data from these two methods, scRNA-seq and scATAC-seq, we aim to construct computational models to study the aberrations in gene regulatory networks.

How do you model gene regulatory networks using single-cell data?

Single-cell sequencing methods produce vast amounts of data from as many as thousands or millions of cells, capturing a plethora of genes and regulatory elements in each of them. The process of gene regulation can be simplified and presented as a graph by utilizing prior biological knowledge on regulatory interactions by connecting different pieces of information within the data, such as regulatory elements to genes by their physical proximity in genome or transcription factors to regulatory elements by the presence of their known binding motifs.

However, the sheer size and complexity of such graphs pose significant computational challenges and traditional statistical methods may struggle to handle the intricacy of the data. This is where artificial neural networks and other advanced machine learning approaches come into play. These methods excel at learning complex patterns from high-dimensional data and can effectively capture non-linear relationships within gene regulatory networks. In this way, we can identify mechanisms associated with development of treatment resistance in cancer cells, which will aid in the development of more efficient treatment strategies in the future.

Want to learn more about the Finnish Center of Excellence in Tumor Genetics Research?
Subscribe to the blog and keep up to date with the latest posts for a peek into the everyday life in cancer research. You can also follow us on Instagram (@tumorgenetics) and Twitter/X! (@CoEinTG)