My Story
My family ran a small village shop when I was a child. Buying and selling kept us going, and my parents taught me numbers and arithmetic early. By age three, I was behind the counter. A customer would hand me money and ask for candy; I would count it out, pack it up, work out the price, and return every last bit of change. I rarely made a mistake. The neighbors loved to tease me about it, and our little shop was often full of laughter.
Then my father’s job took us to another town. One curious neighbor liked to borrow me to mind her shop. People would gather just to watch a child balance the books to the last cent. That early feel for numbers followed me to school. I often earned full marks in math, and I won math competition prizes without extra tutoring or endless practice sheets. I assumed I would keep following that path.
Life, of course, had other plans. When it was time to choose a university major, the claim that “the 21st century belongs to life sciences” won me over. I chose life sciences without hesitation. I was hardly alone: from 1999 to 2007, many of China’s top entrance exam scorers made the same choice. Once classes began, though, the dream met reality. There were endless facts about plants, animals, and anatomy to memorize. I drifted through my days, unsure how to study and even less sure where I belonged. The field I had entered with such hope became what people called a major with bleak prospects. As admission scores fell year after year, I felt the distance between the promise and the reality. That was my first hurdle: I had plenty of passion, but nowhere to put it.
Graduate school at the Chinese Academy of Sciences came as genomics was taking off. My adviser worked in cell biology, but a series of turns led me into evolutionary genomics and then bioinformatics. There, at last, I found my way back to the quantitative thinking I had loved as a child. I learned that an early interest can shape a lifetime of work.
The next wave brought next generation sequencing, whole genome sequencing, RNA sequencing, and proteomics. I rode that wave. I published more than 50 papers in international journals, received national awards and research funding, and became a full professor while still relatively young. Colleagues told me I could simply enjoy what I had built until retirement. For a while, I almost did. But a problem was growing beneath the calm: traditional statistics struggled with the flood of biological data. Breakthroughs became harder to find, and publishing became harder too. I could see the road ahead but could barely move along it. That was my second hurdle: from a position others envied, I could see the ceiling above me.
Then AI began changing every field, including mine. My students showed me the opening. When your students know more than you do about a new subject, the best thing a teacher can do is encourage them and let them pull you forward. I am grateful for those remarkable young people. Our shared enthusiasm and humor have kept us close, and their kind words online still mean a great deal to me. Even so, entering AI from scratch was frightening. I had to rebuild my knowledge from the ground up. I started arriving at the office at five in the morning and leaving at eleven at night, as if I were back in high school preparing for exams. Some nights, after struggling through one more difficult concept, I asked myself whether all this effort at my age was worth it. That was my third hurdle.
Thankfully, the work paid off. We made the transition. Take protein research: we used to extract sequence features by hand. That worked at first, but eventually we hit a wall. With protein language models such as ESM2, a single sequence can yield 5,120 features and capture relationships between distant parts of a protein. This can make protein function predictions more accurate and interpretable, and help models work across species. In synthetic biology, it can also address a pressing industrial need: finding enzymes with better catalytic performance.
I went on to study data science, machine learning, deep learning, and cloud computing systematically. I once earned a top global score on DataCamp, honed my skills in Kaggle competitions, and explored targeted therapy with a Harvard professor. Over time, I learned to connect mathematics, linear algebra, statistics, computer science, software and data engineering, cloud computing, machine learning, bioinformatics, computational biology, genomics, transcriptomics, proteomics, structural biology, immunology, precision oncology, and pharmacology. The longer I study them, the more connections I see. Those connections are what I hope to share in this blog.
I am grateful to live in a time when knowledge is easier to reach than ever. Anyone willing to put in the work can enter a new field.
Growing older is inevitable. Losing your curiosity is not. At any stage of life, the desire to learn can open the door to a new beginning.