Search for: All records

Spermatogenesis is a key developmental process underlying the origination of newly evolved genes. However, rapid cell type–specific transcriptomic divergence of theDrosophilagermline has posed a significant technical barrier for comparative single-cell RNA-sequencing studies. By quantifying a surprisingly strong correlation between species- and cell type–specific divergence in three closely relatedDrosophilaspecies, we apply a statistical procedure to identify a core set of 198 genes that are highly predictive of cell type identity while remaining robust to species-specific differences that span over 25 to 30 My of evolution. We then utilize cell type classifications based on the 198-gene set to show how transcriptional divergence in cell type increases throughout spermatogenic developmental time. After validating these cross-species cell type classifications using RNA fluorescence in situ hybridization and imaging, we then investigate the influence of genome organization on the molecular evolution of spermatogenesis vis-a-vis transcriptional bursting. We first show altering transcriptional burst size contributes to premeiotic transcription and altering bursting frequency contributes to postmeiotic expression. We then report global differences in autosomal vs. X chromosomal transcription may arise in a developmental stage preceding full testis organogenesis by showing evolutionarily conserved decreases in X-linked transcription bursting kinetics in all examined somatic and germline cell types. Finally, we provide evidence supporting the cultivator model of de novo gene origination by demonstrating how the appearance of newly evolved testis-specific transcripts potentially provides short-range regulation of neighboring genes’ transcriptional bursting properties during key stages of spermatogenesis. 

Previous evolutionary models of duplicate gene evolution have overlooked the pivotal role of genome architecture. Here, we show that proximity-based regulatory recruitment by distally duplicated genes is an efficient mechanism for modulating tissue-specific production of preexisting proteins. By leveraging genomic asymmetries, we performed a coexpression analysis onDrosophila melanogastertissue data to show the generality of enhancer capture-divergence (ECD) as a significant evolutionary driver of asymmetric, distally duplicated genes. We use the recently evolved geneHP6/Umbreaas an example of the ECD process. By assaying genome-wide chromosomal conformations in multipleDrosophilaspecies, we show thatHP6/Umbreawas inserted near a preexisting, long-distance three-dimensional genomic interaction. We then use this data to identify a newly found enhancer (FLEE1), buried within the coding region of the highly conserved, essential geneMFS18, that likely neofunctionalizedHP6/Umbrea. Last, we demonstrate ancestral transcriptional coregulation ofHP6/Umbrea’s future insertion site, illustrating how enhancer capture provides a highly evolvable, one-step solution to Ohno’s dilemma. 

We propose a new learning framework that captures the tiered structure of many real-world user-interaction applications, where the users can be divided into two groups based on their different tolerance on exploration risks and should be treated separately. In this setting, we simultaneously maintain two policies π^O and πE: π^O ("O" for "online") interacts with more risk-tolerant users from the first tier and minimizes regret by balancing exploration and exploitation as usual, while π^E ("E" for "exploit") exclusively focuses on exploitation for risk-averse users from the second tier utilizing the data collected so far. An important question is whether such a separation yields advantages over the standard online setting (i.e., π^E=π^O) for the risk-averse users. We individually consider the gap-independent vs. gap-dependent settings. For the former, we prove that the separation is indeed not beneficial from a minimax perspective. For the latter, we show that if choosing Pessimistic Value Iteration as the exploitation algorithm to produce π^E, we can achieve a constant regret for risk-averse users independent of the number of episodes K, which is in sharp contrast to the Ω(logK) regret for any online RL algorithms in the same setting, while the regret of π^O (almost) maintains its online regret optimality and does not need to compromise for the success of π^E.