Abstract:

As humanity evolved, it has significantly advanced in all sciences and fulfilled many challenges. In our opinion, the discoveries in biology and medicine were some of the most impactful; still, many more problems are waiting to be solved. This paper proposes a learning material designed for higher education in computer science that engages a current challenge, the lack of an HIV vaccine and cure.

The learning material is built around existing research in bioinformatics, HIV vaccine design, and machine learning. The students are presented with real experimental data from HIV vaccine research and an algorithm based on neuronal networks that process this data. The students' task is to improve the existing algorithm or find brand new ideas after gaining inspiration from the current material. As the students try to solve this challenge, they are making their first steps from the realm of learning into researching. The material is well suited for a course on machine learning or bioinformatics.

In our opinion, the material is a complex one that pushes the students out of their comfort zones and stimulates their capabilities; however, because of the involved difficulty, the teachers will need to be a bit more indulgent than usual. The students can acquire the knowledge in more ways; usual classroom teaching could be employed, but case-based and problem-based learning might work even better with how we designed this material. In case-based and problem-based learning, as opposed to classical learning, the student needs to do self-research while the teacher makes minimal interventions. Gamification is another potential strategy; the algorithm is evaluated using the data and statistical metrics; therefore, the students could be split into groups and compete to create better versions that record higher metrics. Last, this material could serve as an excellent inspiration for a PhD thesis.

This work complements the other two that we created previously. Those covered the application of neural networks to two computer vision tasks in medicine: the processing of colposcopy images and of lungs x-ray images. Therefore all three learning materials, which are publicly available, can be combined to form a more comprehensive collection. All materials can be thought and accessed online, being based on Jupyter notebooks.

Keywords:

Bioinformatics, HIV, vaccine, antibodies, deep learning, neuronal network, machine learning, Jupyter notebook.