In between his works at his A.I. company, his biotech company, and numerous public lectures, he taught an Integrated Science class every fall at Itiwo Schools. Dr. Asanmu was a polymath for sure. One moment he is seen sitting down with bishop-philosophers discussing the intricacies of Thomism and Augustinianism, the next minute he is seen across the street with hardcore biologists musing about telomere biology and how to solve the problem of aging.
Dr. Asanmu was one of those solid teachers. He always looked as if he grew up in a rush, intellectually; as though growing up was a race. He spoke softly, yet firmly. Always gesticulated, always, and he never gets tired. For some reason, he doesn’t. His analogies are like dark roasted coffee; they will keep you awake!
Today he promised to talk about a computational biochemistry project that was ongoing in his biotech startup. I am always happy to welcome him to our little school. We attended Itiwo over twenty years ago for junior secondary school, he was one of those shy guys always mistaken to be arrogant, and the moment you get close, those assumptions melt. He had a B.S. in Mathematics, a PhD in Biochemistry, and somehow found himself in the startup world.
Today, the kids were excited as always to see the self-effacing Asanmu teach.
He had introduced them to cancer biology in the previous class. Today, he began the class with his classic soft-firm voice, and an awkward sentence.
“It started with blood in the urine, then he was getting unusually tired. He would play tennis with us after work but for about a week now, the first two bursts of sprinting and he was done. Done. Dead Tired. He figured he needed to see the Doctor.”
Asanmu paced the classroom gently as though he walked on a glass, his eyes hopped and danced around the red stained wooden floor, gesticulating. He continued, “he went to the doctor and the pretty lady told him, very, very politely that he was as good as dead both kidneys were gone, and now the cancer cells had carried the riotsto the spinal cord, ravaging everything in their path, traveling ruthlessly.”
As he told the story, I swear, if a pin dropped in the classroom it would wake a dead man. It was the story of his old buddy who died eight years ago from kidney cancer. He told the students that he would call it kidney cancer, even though the technical name is renal cell carcinoma. His friend, Bobby, was diagnosed with stage IV kidney cancer, and he died a few months afterwards.
Shortly after Bobby’s death, he told the students, he launched a cancer biology startup and was going to discuss one of the main projects today.
“First the significance,”he said.
“One of the reasons that made the quick death of my old friend Bobby from cancer almost certain was because it was diagnosed late. We know that with such late diagnosis, we are talking only 2% survival rate, few months tops.”
“And why was he diagnosed so late?” Gbemi, a young lad in the class asked.
“Because cancers in general are asymptomatic, unless they are in very vital parts of the body, say the brain. Symptoms only tend to ensue especially when the disease has metastasized to other parts of the body. That is, when it had broken bridges and damaged a lot of pipes. On the other hand, you get an early diagnosis and the chance of survival is up there in the sky.”
Asanmu continued and explained to the student that the main, current method of diagnosis of kidney cancer is via biopsy. A student asked what a biopsy was, and he replied, smiling back at the young chap.
“It is when the surgeon takes some of your kidney tissues to see if it’s indeed cancerous.”
“And you have to do that each time to get diagnosed?” the student retorted.
“Yes, something like that,” Asanmu said.
He explained that diagnosis via biopsy is riddled with sampling errors and also invasive. All of which come at a huge psychological and financial costs to the patients.
“Next, let’s talk metabolism,” he said.
Asanmu went ahead and explained that when food is ingested into our bodies, they are broken down into bits and pieces. This breaking down of nutrients releases energy, and such energy is used to do work, all sorts of work, like growing up. This metabolism, he said, is the combination of all those chemical reactions that therefore keeps us alive. He likened metabolism to a ‘huge, huge carnival’. Even though it appears as though everyone is doing their own thing, the different choreographies mesh together to give us a lively carnival. This he called an ‘emergent property.’
“Back to cancer and symptoms,” he said calmly, as he fogged his clear glasses with his breath and wiped the lenses clean with a white handkerchief.
“Cancer as we talked about in our last class, is when cell growth goes rouge and sometimes spreads mercilessly to other body parts.”
He paused for a couple seconds, put on his glasses and continued.
“Cancer is like an intruder in our carnival. The problem is, this disruption is not always obvious, especially if it is localized in a non-vital organ, so it continues to spread. And by the time it becomes obvious, it is often too late, as many parts of the carnival are already affected, and this could well be the end of our carnival. Now, if we are on the planning committee of the next, say, Rio Carnival in Brazil, what should we do?
“Check for the intruders before they mess up the carnival,” Amaka, the dotty girl from the East replied quickly.
“Yes, yes!” Asanmu shrieked, gleefully.
Now holding Amaka in the hand, and helping her from her seat to the front of the class.
“What Amaka and I will do is to set up a sleek system that will routinely check for intruders up close duringthe course of the carnival, especially if we are in a high crime environment. And one more thing, we don’t want to interrupt their performances, do we Amaka?
“No, no.” She replied as she shook her head intensely.
“So, what do we do? We want to check for intruders, and we don’t want to risk interrupting performances, what do we do?”
“Hmmm, somehow check if dancers are smiling or frowning?” Amaka replied.
“Haaa, beautiful! Class, we have a plan, we have a great plan. Let’s get back to science”
Asanmu guides young Amaka to her seat as the entire class watched on smiling, giggling, teasing Amaka.
“Similarly,” he continued, “If you are a biochemist that studies cancer metabolism, you want to develop tests to routinely check for any indication of cancer that will be recommended to people who are at risk of certain cancers; or for folks physicians think might have cancer. Also, like we agreed, we don’t want to be going into the patient each time we do this, that’s invasive. We want to detect an indication of something that might be wrong, very far from the scene, with smiley faces or frowning faces like Amaka suggested.”
“It is important to state explicitly at this point that what cancer disrupts is our body metabolism. A choreographer or a dancer is a metabolite. A particular kind of choreography by a group of choreographers is a metabolic pathway, and finally our entire carnival is The Metabolism. And, class, all we want to see is if our choreographers are frowning or smiling as a proxy for the carnival’s health.”
Asanmu watched the class like a hawk to see if the analogies sank in. And everything he saw was very good. There were smiles and cheers, and he continued.
“Next, what we want is to collect urine samples from kidney cancer patients and then see what the metabolite levels are compared to the healthy controls. This is a perfect plan, believe me, especially given the close relationship between the urine and the kidney.”
Asanmu went ahead and explained that they used two different analytical chemistry platforms for the experiment and that not many people do this in the field. The reason for this he said is to ‘capture more choreographers.’ One he called ‘NMR’, another he called ‘Mass Spec’. He said these platforms helps detect and identify the different types of metabolites in the urine, and quantify their levels. He explained the importance of controlled experiments to the kids, that they have also collected urine of several healthy individuals for the basis of comparison. He said, ‘after many, many steps’ they end up with lots of data, and the next step is to make sense of the data using machine learning.
Immediately one of the kids asked what machine learning was.
“Machine learning is the way machine learns, which is conceptually similar to the way humans learn.”
“Can anyone tell me here how we learn?”
Amaka quickly interjected, “we learn by reading for our class.”
“Great!” Asanmu responded. “Any other person?”
“By watching out for things in our environment,” another added.
“These are great answers. And you get the gist. We are learning all the time, whether for a class or in our social environment. And the way we learn is to look for patterns. For example, our eyes, ears, skin, nose are super-duper sensors. We use them to gather information about the environment and this information are processed in our brain in order to detect patterns. Once these patterns are learnt, we have got what is needed to live reasonably well in our social environment. Come to think of it, there is a reason we are not constantly getting bumped into trees, or getting our fingers burnt. Indeed, folks who have defects in their brain, folks who have lost the ability to learn certain patterns, behave quite differently.”
“Similarly,” he continued, “what we are trying to do with this project is to mimic the process I have just explained. In this case we have captured the metabolites information from each subject in the study using our analytical chemistry platforms. Next, we want to see if we can program a computer to learn which subject is healthy and which subject has kidney cancer just from the metabolites information.”
“And that will be predicting if a carnival is doing fine or infiltrated, just from the faces of the choreographers?” A student asked.
“Yup!” Asanmu replied happily.
“And how will the computer learn this?” another student asked.
He went further to explain that there is a biological assumption here which is quite solid, that there is a mathematical relationship, an association between the levels of certain metabolites and whether a subject is healthy or has tumor. A relationship between a cancerous or healthy carnival, and the frowning or smiling faces of certain choreographers.
With that, we can make a machine learn by tagging the metabolite profile of a subject with a label. This label is binary, either healthy or kidney cancer. The next step will be to split this dataset into two. For the first split, you make your machine learn the mathematical relationship between metabolites level and labels. For the second set, you remove the label, and make the machine predict what the labels were, just from the metabolites level, to test if the machine had indeed learnt the mathematical relationship.
“This is very similar to when you prepare for an exam,” he said.
“If you learn all the examples of the class, and if the tutor gives you the exact examples from the class notes in the exams and you pass them all, it’s tricky to establish that you have learnt. You might just have done a great job memorizing the examples. The best way to know you have learnt is to give you a completely different set of questions that test the same concepts. If you have not appropriately learnt the patterns, you will be caught in no time.”
“When we did all these,” Asanmu continued, “we were glad to find a small set of metabolites in the urine that were able to predict kidney cancer accurately. We call these the biomarkers.”
“That’s great!” One of the students yelled.
“Now people can get their urine tested for kidney cancer in the clinic, at home?” another asked
“Not that fast. This first stage is called the discovery phase. The urine biomarkers need to pass even more tests. From our class-exams analogy, that will be a huge load of different questions testing the same concept.”
He paused for a second and began to roll down his sleeves, a signature move signaling the end of the class.
“The next stage will be a validation stage, where thousands of individuals that are even more different, in the context of age, gender, dietary habits, are subjected to kidney cancer predictions using the urine biomarkers. And hopefully when we are done, the Bobbys of this world, won’t have to have their carnivals untimely disrupted just because we couldn’t capture the smiling and frowning faces of the choreographers and dancers.”
Asanmu fielded a few questions from the gleeful students, finally buttoned up his sleeves and called it a day.