Bayesian analysis of longitudinal RB-TnSeq resolves the fitness seascape in fluctuating environments

Technical Summary: Bayesian Analysis of Longitudinal RB-TnSeq

1. Problem Statement

In natural ecosystems, environmental conditions are rarely static; they fluctuate temporally. Consequently, the fitness of a genotype is not a single value but a time-dependent function. While genome-scale fitness landscapes can be mapped in stable environments using methods like RB-TnSeq (Randomly Barcoded Transposon Sequencing), measuring time-dependent fitness effects remains a significant challenge. Traditional methods often struggle with:

• Noise: Stochastic fluctuations in mutant abundances over time.
• Complexity: The difficulty of distinguishing between transient fitness changes and long-term evolutionary trajectories.
• Predictive Gap: The inability to link short-term selection coefficients to long-term adaptive outcomes in fluctuating regimes.

2. Methodology

The authors address these challenges through a multi-layered computational and experimental approach:

• Bayesian Multilevel Framework: To process longitudinal RB-TnSeq data, the authors developed a Bayesian hierarchical model. This framework treats mutant abundance trajectories as noisy signals, using a multilevel structure to stabilize estimates. This allows for the extraction of time-resolved selection rates while providing formal, interpretable uncertainty (credible intervals) for each mutant at each time point.
• Experimental Model: They utilized Escherichia coli subjected to a "feast-famine" starvation regime. This creates a predictable but fluctuating environment where the selective pressures shift between rapid growth (feast) and survival/maintenance (famine).
• Fisher’s Geometric "Seascape" Model: To simplify the high-dimensional fitness data, the authors employed a one-dimensional compression technique. They mapped mutants onto a latent axis (a "seascape") based on Fisher’s Geometric Model, which characterizes how mutants navigate trade-offs between different environmental optima.
• Validation via Long-term Evolution: The model’s predictive power was tested by comparing short-term longitudinal fitness estimates against the actual mutational targets identified in an extended, long-term evolution experiment under the same conditions.

3. Key Contributions

• New Computational Toolset: A robust Bayesian framework specifically designed for longitudinal barcoded sequencing data, enabling the quantification of temporal selection.
• Dimensionality Reduction of Fitness: The "seascape" model provides a way to compress complex, time-varying genome-wide data into a single, interpretable axis of evolutionary strategy.
• Integration of Scales: The study successfully bridges the gap between short-term competitive fitness (how a mutant performs in a single cycle) and long-term adaptive evolution (which mutants actually persist and dominate over many generations).

4. Results

• Identification of Fitness Trajectories: The analysis revealed that mutants follow distinct trajectories. Many mutants exhibited "shared molecular strategies," meaning different genes contributed to fitness in similar ways depending on the growth phase (growth vs. starvation).
• Antagonistic Pleiotropy and Constraints: The study identified significant trade-offs. Specifically, they found that initial growth fitness is a dominant constraint. Mutants that were deleterious during the initial growth phase were unable to be "rescued" by later advantages during the famine phase, regardless of how beneficial they were under stress.
• The Generalist-Specialist Axis: The "seascape" model successfully ordered mutants along a latent axis that represents the trade-off between being a "generalist" (stable fitness across both phases) and a "specialist" (high fitness in one phase but low in the other).
• Predictive Accuracy: The longitudinal estimates and the inferred seascape were highly predictive of which mutants would emerge as adaptive targets in long-term evolution experiments, confirming that short-term temporal dynamics dictate long-term evolutionary trajectories.

5. Significance

This work represents a major step forward in evolutionary systems biology. By moving beyond "static" fitness landscapes, the authors provide a mathematical and experimental blueprint for studying how organisms evolve in the real, fluctuating world. The ability to resolve the "fitness seascape" allows researchers to quantify the fundamental constraints of evolution—such as antagonistic pleiotropy—and predict how populations will respond to environmental instability, which has profound implications for understanding microbial adaptation, antibiotic resistance, and ecosystem dynamics.