- Cancer Care Now Runs on Data Infrastructure

Why the Future of Oncology Depends on More Than Medicine
By David Hutchful
When most people think about advances in cancer care, they think about new drugs, new therapies, or breakthrough discoveries in biology. But increasingly, one of the most important determinants of progress in oncology is something far less visible: data infrastructure.
Modern cancer care generates enormous amounts of information. Pathology reports, imaging, genomic sequencing data, treatment histories, laboratory results, adverse events, survival outcomes, and physician notes all contribute to a patient’s cancer story. Yet in many healthcare systems around the world, this information remains fragmented across disconnected systems.

Photo Credit: Gemini AI/Google
A patient may receive imaging in one facility, surgery in another, chemotherapy in a third, and genomic testing elsewhere entirely. The result is that clinicians often have only partial visibility into the patient journey, while researchers struggle to assemble high-quality datasets capable of generating meaningful scientific insight.
This fragmentation is not simply an operational inconvenience. It directly affects the future of precision oncology.
At Yemaachi, we think about cancer not only as a biological challenge, but also as a data and systems challenge. In many ways, modern oncology increasingly resembles other data-intensive industries where infrastructure determines what becomes possible.
Just as financial systems depend on trusted digital infrastructure to move money reliably, precision medicine depends on trusted biomedical infrastructure to move information reliably.
The quality of that infrastructure matters enormously.
A genomic sequence without clinical context has limited value. A pathology result without longitudinal follow-up limits research usefulness. Even small inconsistencies in how patient information is recorded can reduce the reliability of downstream analyses.
This is why oncology increasingly depends on structured, interoperable, and longitudinal data systems.
Structured data means information is captured consistently. Interoperable data means systems can communicate with one another. Longitudinal data means researchers and clinicians can understand how disease evolves over time.
Together, these capabilities create the foundation for several important advances:
- More accurate biomarker discovery
- Better patient stratification
- Improved treatment selection
- Faster clinical trial recruitment
- Better understanding of treatment response across populations
- More effective development of future therapies
One important misconception is that data infrastructure is primarily a technical concern. In reality, it is deeply clinical.
Every physician note, pathology annotation, biospecimen label, staging update, and treatment outcome contributes to the integrity of the broader system.
The future usefulness of cancer datasets often depends on decisions made during routine clinical care.
For example, incomplete staging information may later limit researchers’ ability to correlate genomic findings with disease progression. Missing treatment timing data may reduce the quality of survival analyses. Inconsistent terminology may complicate machine learning workflows designed to identify patterns across thousands of patients.
Small documentation decisions can have long scientific consequences.
This becomes even more important in the era of artificial intelligence (AI).
AI systems are often discussed as though they are independent technologies. In reality, AI systems inherit the strengths and weaknesses of the data used to train them.
If datasets are fragmented, biased, incomplete, or poorly standardized, AI models will reflect those limitations.
Conversely, high-quality clinical and molecular datasets create the conditions for more reliable and clinically useful AI tools.
This is one reason why globally diverse cancer datasets are becoming increasingly important.
Historically, many genomic reference databases have disproportionately represented patients from a limited number of populations. As precision medicine advances, there is growing recognition that broader population representation improves the scientific robustness of cancer research for everyone.
The goal is not diversity for its own sake. The goal is scientific completeness.
At Yemaachi, we believe building high-quality biomedical infrastructure is foundational to innovations in oncology. The future of cancer research will not depend solely on discovering new molecules. It will also depend on building systems capable of learning continuously from every patient interaction.
This is what learning health systems aspire to achieve: clinical care and research improving one another in a continuous feedback loop.
In this future, every patient encounter contributes, securely and ethically, to improving future care.
And behind that future lies infrastructure.
Not only biological infrastructure.
But digital infrastructure, computational infrastructure, and data infrastructure operating together.
The future of oncology will belong not only to institutions that generate data, but to those capable of organizing, interpreting, and learning from it responsibly.©
- FFPE Tissue: A Goldmine for Cancer Genomics Research

By Joyce Ngoi, Max Nwuba & Elikem Kisser
Cancer samples collected in the form of Formalin-fixed paraffin-embedded (FFPE) tissue are most often the primary source for the characterisation of tumours. FFPE specimens are routinely obtained from patients during biopsies or surgical resections and preserved for long-term clinical diagnosis and research, making FFPE the gold standard for tissue preservation. The preservation process involves the use of fixatives and embedding media such as formalin and paraffin, respectively, to maintain the tissue morphology and cellular structure as close to the natural state as possible for long-term storage. This makes FFPE tissues extremely useful for several lab processes, such as immunohistochemistry, histopathology, and molecular analysis, and thus, a valuable resource for cancer research. When the tissue is preserved this way, the DNA and other contents are preserved. This means that researchers can extract genetic material from tumour tissues long after the sample was collected.
The Double-Edged Nature of FFPEWhile formalin is excellent at preserving tissue structure, its chemical mechanism can also damage nucleic acids. Formalin creates unnatural chemical links: DNA-DNA, DNA-Protein, and Protein-Protein crosslinks, which make it difficult to extract and prepare DNA for sequencing. Beyond inducing crosslinking, formalin fixation also triggers various forms of DNA damage, including cytosine deamination, oxidative stress, nicks, gaps, and abasic sites. Coupled with the often limited amount of tissue available from diagnostic biopsies, these changes make it challenging to recover sufficient quantities of high-quality DNA for downstream applications such as whole-genome sequencing.
Formalin fixation therefore presents a paradox. It preserves tissue beautifully for research and long-term storage, yet this same process can complicate the genetic analysis researchers hope to perform years later. A helpful analogy is if one orders a 500-piece puzzle and receives a 50,000-piece puzzle instead. The picture is still there, but now it’s been shattered into thousands of tiny fragments, and the effort required to reconstruct the original image increases dramatically.
“It’s an excitingly frustrating challenge”
From Tissue to Genetic Data

Extracting genetic information from FFPE samples for cancer research is neither straightforward nor as simple as fresh tissues. Extreme precaution at different analytical stages needs to be observed when working with FFPE samples to mitigate frequently occurring issues, which could potentially lead to misinterpretation of mutations in the genomic data. Despite all of these challenges, there has been significant advancement in finding solutions for whole-genome sequencing of FFPE samples.
The first step is to extract DNA from FFPE tumour tissues, which are usually stored as thin sections embedded on slides. The extraction procedure has to be carefully optimised and curated for efficient and successful DNA isolation. This basically includes a well-controlled process that prevents further damage, reverses the formalin-induced cross-linking, while ensuring sufficient DNA yields are obtained. In addition, automation of the extraction workflow standardises procedures, enhancing accuracy, efficiency and productivity. Quality analysis of the extracted DNA is crucial for selecting samples that are suitable for sequencing. The analysis determines the degree of damage in FFPE-derived DNA and identifies metrics that correlate with DNA quality. This allows sequencing outcomes to be predicted and unsuitable samples to be identified early, enabling their exclusion from sequencing to reduce costs and prevent the generation of low-quality, unusable data.
The DNA extracted from FFPE tissue is further prepared for whole genome sequencing (WGS) for a comprehensive overview of the genome. The human genome contains about 3.2 billion base pairs (base pairs are the fundamental building blocks of DNA). Since this is too much for the sequencing instruments to read all at once, the genome is first divided into many short fragments called “libraries”. These libraries are designed with modifications that allow the DNA fragments to be read properly by the sequencer. However, because the FFPE tissue contains damaged DNA, preparing it for sequencing is challenging and requires a well-designed library preparation and sequencing protocol to obtain meaningful data. Only samples that pass rigorous quality control checks for DNA quality and quantity proceed to library preparation. Every step of library preparation must be extremely precise to target, repair, and amplify the remaining usable DNA fragments. This is because all the mechanisms that cause DNA damage lead to loss of sequence information, and the amount of usable DNA in the sample should therefore be maximised.
The libraries undergo a quality control step to verify that fragment sizes are within the expected range and that sufficient material is available to generate high volumes of high-quality sequencing data. Sequencing is then performed on a high-throughput sequencing platform like the Illumina NovaSeqX Plus. Within the sequencer, each short DNA fragment is read billions of times over. This repeated “reading” ensures that the digital DNA sequence is accurate and the mutations observed are truly present in the tumour DNA.
What follows are downstream computational processes that help us understand the full picture of genetic changes present in the tumour. The information gathered from the sequencing process is then analysed using high-performance computing systems. Billions of tiny pieces of genetic code are put back together, like rebuilding a massive puzzle, to create a complete picture of the patient’s genetic information.All of these intricate molecular techniques help us unlock valuable genomic information in the field of cancer research, transforming years-old FFPE blocks into previously unseen genomic data from the African continent.