Nope, not a chance that we won't need bench work.

You are assuming that the data all exist & the AI is there to look for new connections. We had this same conversation when the human genome was sequenced - we can solve all human diseases! Design genomics medicine! etc.

It's been 20+ years since the first human genome draft was published. In that time, we can't even predict mRNA splice sites with 100% accuracy. We've discovered the importance of microRNAs, long-distance regulatory regions, and Transcriptional elements in genome evolution. These were all lumped in to a category of "junk DNA" since we didn't know what they were and/or their functions.

In 5 years, we'll have another big technology jump and we'll look back at this AI enthusiasm and wonder why we were so excited when we simply do not have the data for AI to be useful.

In my opinion, we need to reach certainty not probability. The wet lab never gets close, and the dry lab helps us define near-to-real hypotheses.

Shabnom Momtaz,

While the rapid advancements in AI and bioinformatics are indeed remarkable, the notion that dry labs might replace wet labs entirely is not plausible. Wet lab experiments are indispensable, as they provide empirical validation and uncover complex biological phenomena that cannot be fully captured by theoretical models or simulations. AI-driven tools like AlphaFold have revolutionized the prediction of protein structures, offering insights that were previously time-intensive to achieve. However, such tools generate theoretical results based on existing datasets, and their predictions often require validation through in vitro and in vivo experiments. These experiments are essential to account for the dynamic and multifaceted nature of biological systems, which often behave unpredictably outside computational models. Moreover, the reproducibility of findings can vary significantly across in silico, in vitro, and in vivo platforms. Dry labs are invaluable for narrowing down hypotheses and optimizing experimental approaches, but wet labs remain the cornerstone of scientific discovery and innovation. Together, dry and wet labs form a complementary partnership that enhances research efficiency and precision, rather than one replacing the other. This synergy ensures that while bioinformatics and AI refine and guide research pathways, the fundamental truths of biology are still revealed through direct experimentation.

I would like to clarify an important point I have read several times. AI is a subset of bioinformatics and not a separate tool. Bioinformatics converts biological questions into computationally manageable data and relies on predictions based on statistics. At the end we are dealing with biological questions and all what bioinformatics does is prediction. Can we, human, trust a treatment just from predictions? GXP will never accept it and biology in proper term is life science. Without wet lab it would be impossible to check the hypothesis/prediction. In addition, we are still far to use only deep learning for accurate output. These algorithms require large, high-quality, and diverse datasets for accurate predictions. Our current knowledge of biological systems, especially complex regulatory networks, is insufficient for such training. Just imagine the regulatory network in each leaving organism, we might know less than 5%, how would we train data based on poor knowledge? We may end with a very sophisticated prediction but not realistic.

Thank you so much Loubna Youssar Bhaskar Chatterjee Isaac Karimi Katie A S Burnette for your valuable answer. I appreciate your answers.