Success Story: AI-Powered Pathogen Detection

Algorithm Development & Future AI Goals: We developed a custom method for pathogen detection that proved to be significantly more accurate than methods used in current literature (e.g., https://doi.org/10.3390/ijms231911150). We used the Plasmodium data of the aforementioned study for validation due to its similarity to Babesia. Crucially, the high-performance CPU/GPU nodes and large RAM resources of KISSKI are essential for our current work in processing vast metagenomic collections. Looking ahead, we are highly interested in exploring how AI-driven architectures, such as BERT, could be utilized to better identify relevant pathogenic patterns within our Long-Read sequencing data. While we are still in the very early conceptual stages of this approach and have yet to realize its implementation, we see great potential in these technologies and would welcome any future guidance or expertise KISSKI could provide in this field.

Case Study (Ötzi the Iceman): We analyzed the metagenomic data of the “Iceman” (https://doi.org/10.1038/ncomms1701). Our analysis identified a “false assignment” regarding Borrelia in the original publication—a finding later corroborated by Prof. Albert Zink, who acknowledged the error. However, we successfully detected traces of other pathogenic agents metagenomically.

Clinical Relevance: Our work focuses heavily on Babesia and Bartonella, aligning with research suggesting these pathogens as potential causes for ME/CFS, which affects 300,000 to 500,000 people in Germany. A key study detected these specific pathogens in nearly half of the affected individuals (https://doi.org/10.3390/pathogens15010002).

Non-Profit Commitment & Future Research

To further our research, I have personally acquired a MinION sequencer and am establishing informal cooperations with specialized laboratories to handle infectious samples. It is important to emphasize that our project is a non-profit self-help initiative. As a medical professional, my goal is to bridge high-level bioinformatic analysis with clinical research to help people who are often left without answers. We operate entirely without commercial interest or external funding; therefore, the continued free-of-charge access to KISSKI is our only possibility to sustain this vital work for the community