Search

The rise of high-throughput experiments has transformed how scientists approach biological questions. The ubiquity of large-scale assays that can test thousands of samples in a day has necessitated the development of new computational approaches to interpret this data. Among these tools, machine learning approaches are increasingly being utilized due to their ability to infer complex nonlinear patterns from high-dimensional data. Despite their effectiveness, machine learning (and in particular deep learning) approaches are not always accessible or easy to implement for those with limited computational expertise. Here we present PARROT, a general framework for training and applying deep learning-based predictors on large protein datasets. Using an internal recurrent neural network architecture, PARROT is capable of tackling both classification and regression tasks while only requiring raw protein sequences as input. We showcase the potential uses of PARROT on three diverse machine learning tasks: predicting phosphorylation sites, predicting transcriptional activation function of peptides generated by high-throughput reporter assays, and predicting the fibrillization propensity of amyloid beta with data generated by deep mutational scanning. Through these examples, we demonstrate that PARROT is easy to use, performs comparably to state-of-the-art computational tools, and is applicable for a wide array of biological problems.

It is increasingly appreciated that intracellular pH changes are important biological signals. This motivates the elucidation of molecular mechanisms of pH-sensing. We determined that a nucleocytoplasmic pH oscillation was required for the transcriptional response to carbon starvation inS. cerevisiae. The SWI/SNF chromatin remodeling complex is a key mediator of this transcriptional response. We found that a glutamine-rich low complexity sequence (QLC) in theSNF5subunit of this complex, and histidines within this sequence, were required for efficient transcriptional reprogramming during carbon starvation. Furthermore, theSNF5QLC mediated pH-dependent recruitment of SWI/SNF to a model promoterin vitro.Simulations showed that protonation of histidines within theSNF5QLC lead to conformational expansion, providing a potential biophysical mechanism for regulation of these interactions. Together, our results indicate that that pH changes are a second messenger for transcriptional reprogramming during carbon starvation, and that theSNF5QLC acts as a pH-sensor.

Many postdoctoral researchers apply for faculty positions knowing relatively little about the hiring process or what is needed to secure a job offer. To address this lack of knowledge about the hiring process we conducted a survey of applicants for faculty positions: the survey ran between May 2018 and May 2019, and received 317 responses. We analyzed the responses to explore the interplay between various scholarly metrics and hiring outcomes. We concluded that, above a certain threshold, the benchmarks traditionally used to measure research success – including funding, number of publications or journals published in – were unable to completely differentiate applicants with and without job offers. Respondents also reported that the hiring process was unnecessarily stressful, time-consuming, and lacking in feedback, irrespective of outcome. Our findings suggest that there is considerable scope to improve the transparency of the hiring process.

Phase transitions of linear multivalent proteins control the reversible formation of many intracellular membraneless bodies. Specific non-covalent crosslinks involving domains/motifs lead to system-spanning networks referred to as gels. Gelation transitions can occur with or without phase separation. In gelation driven by phase separation multivalent proteins and their ligands condense into dense droplets, and gels form within droplets. System spanning networks can also form without a condensation or demixing of proteins into droplets. Gelation driven by phase separation requires lower protein concentrations, and seems to be the biologically preferred mechanism for forming membraneless bodies. Here, we use coarse-grained computer simulations and the theory of associative polymers to uncover the physical properties of intrinsically disordered linkers that determine the extent to which gelation of linear multivalent proteins is driven by phase separation. Our findings are relevant for understanding how sequence-encoded information in disordered linkers influences phase transitions of multivalent proteins., eLife digest Our cells contain a variety of structures called organelles that perform specific roles within a cell. Some organelles are surrounded by a membrane, while others float inside the cell as spherical droplets made of proteins. These proteins contain several sticky regions, which are connected by flexible linker proteins. It is thought that the level of stickiness and the number of sticky regions, or domains, determine whether a protein will form a membraneless organelle. Often, proteins with similar sticky domains have different linkers, and until now, it was assumed that the linkers do not have any other purpose than stringing the domains together. To test this further, Harmon et al. used a combination of computer simulations and physics-based theory. In these simulations, the domains were kept the same, but the properties of linkers were changed to see if this would influence how the membraneless organelles are formed. The results showed that depending on the physical properties of the linkers, the proteins could huddle together and form dense spherical gel-like droplets similar to the membraneless organelles, or form open non-spherical gels. When the linkers were short, the proteins do not easily form droplets. Linkers that were sufficiently long but too bulky, lead to non-spherical gels. Compact linkers, however, enabled proteins to huddle and form spherical gels. The spherical droplet-spanning gels required much less protein compared to the open non-spherical gels. This suggests that proteins important for forming membraneless organelles can be distinguished from those that are not based on the properties of their linkers – even when their domains are similar. These findings further scientists’ knowledge of how specific types of proteins form membraneless organelles and will help to understand how membraneless organelles control many key aspects of how a cell works.