Gašper Tkačik

The paper shows that irreversible differentiation programs are common due to their association with stability and robustness, not direct selection.

The paper shows that cells specify fate within an optimal window of positional information during zebrafish neurulation, where Shh gradient information peaks early and declines despite increasing signal activity.

The paper identifies intrinsic population dynamics in the superior colliculus that modulate visuospatial attention and enhance visual responses during a detection task.

The paper shows that the scaling exponent of hippocampal activity during sleep/rest predicts spatial memory performance independently of reactivation.

The paper derives an optimality principle for gene regulatory architectures using population genetics and information theory, showing how evolutionary selection optimizes genotype-phenotype maps.

This study uses a biophysical model to show how regulatory architectures evolve to balance crosstalk and evolvability by optimizing short, weak binding sites and synergistic activation.

The paper shows that basal gene expression in bacteria is not inert but consists of stochastic pulses that enhance fitness by accelerating growth and reducing lag times during environmental transitions.

This study shows that the hippocampus maintains a hierarchical representation structure during learning but modifies lower-level representations to improve context decoding and learning efficiency.

This paper shows chromatin reduces regulatory errors in gene expression by silencing off-target genes and enabling scalable gene regulation.

The paper develops a mechanistic model of bacterial promoters to explore how molecular complexity affects evolutionary predictions, finding that simplifying mechanisms does not compromise accuracy.

The paper introduces the quantitative omnigenic model (QOM) to improve GWAS interpretability by integrating regulatory network knowledge, showing it can predict gene expression with fewer parameters and extract causal effect chains.

The paper reveals structured cell-to-cell interactions in hippocampal CA1 that optimize spatial coding in familiar and novel environments through statistical modeling and maximum entropy frameworks.

The paper shows that non-specific TF-DNA binding in bacteria reduces population growth across environments due to low specificity, TF cooperativity, and TF-to-DNA ratios.

This paper introduces a framework using multi-agent reinforcement learning to train and infer neural network functions in systems neuroscience.

The paper develops a model of population coding in the visual cortex that optimizes perceptual inference while minimizing activity cost, reproducing attentional modulation properties and generating hypotheses about top-down feedback and noise correlations.

This paper unifies normative theories and statistical inference in a Bayesian framework for biological systems analysis, demonstrating its application to neural data.

This paper introduces a probabilistic neural model inspired by sparse random connectivity that efficiently estimates spiking patterns in neural circuits and outperforms existing models.

The paper investigates drug interactions between translation-inhibiting antibiotics by modeling them as translation bottlenecks and shows how kinetic factors and continuous epistasis influence these interactions.

This paper introduces a method to separate intrinsic neuron interactions from extrinsic correlations in sensory neural networks using mean-field corrections, validated on retinal recordings.

The paper quantifies how extracellular information is encoded in the dynamic nuclear localization of transcription factors in yeast, showing that collective dynamics, not individual factors, represent environmental states.

The paper develops a unified framework to compare and extend theories of efficient, predictive, and sparse coding in sensory neurons, showing how different coding objectives affect neural responses.

The paper shows that a single type of retinal ganglion cell encodes two different visual features depending on stimulus context.