Introductory bioinformatics courses can be challenging to teach. Students with a biological background may have never encountered computer science, and computer science students are likely to have minimal knowledge of biology. To improve learning, we implemented a flipped spaced-repetition course. We repeated the topics through various activities across different days while applying an unusually high number of examinations. The examinations were synergistic with the flipped classroom, encouraging reading and watching recorded lectures before in-person discussions. Additionally, they helped us structure and assess laboratory practicals. We analyzed grades, pass rates, student satisfaction, and student comments qualitatively and quantitatively over 7 years of the course, documenting progress as well as the effect of disruptions such as COVID-19 and changes in teaching staff. We share our results and insights into the opportunities and challenges of this pedagogical approach. An open online version of this course is freely provided for students and teachers.

We show that a small biotin-binding RNA aptamer that folds into a pseudoknot structure acts as a substrate for bacterial RNase P RNA (RPR) with and without the RNase P C5 protein. Cleavage in the single-stranded region in loop 1 was shown to depend on the presence of a RCCA-motif at the 3’ end of the substrate. The nucleobase and the 2’hydroxyl at the position immediately 5’ of the cleavage site contribute to both cleavage efficiency and site selection, where C at this position induces significant cleavage at an alternative site, one base upstream of the main cleavage site. The frequencies of cleavage at these two sites and Mg2+ binding change upon altering the structural topology in the vicinity of the cleavage site as well as by replacing Mg2+ with other divalent metal ions. Modelling studies of RPR in complex with the pseudoknot substrates suggest alternative structural topologies for cleavage at the main and the alternative site and a shift in positioning of Mg2+ that activates the H2O nucleophile. Together, our data are consistent with a model where the organization of the active site structure and positioning of Mg2+ is influenced by the identities of residues at and in the vicinity of the site of cleavage.

The revolution in cryo-electron microscopy has resulted in unprecedented power to resolve large macromolecular complexes including viruses. Many methods exist to explain density corresponding to proteins and thus entire protein capsids have been solved at the all-atom level. However methods for nucleic acids lag behind, and no all-atom viral double-stranded DNA genomes have been published at all. We here present a method which exploits the spiral winding patterns of DNA in icosahedral capsids. The method quickly generates shells of DNA wound in user-specified, idealized spherical or cylindrical spirals. For transition regions, the method allows guided semiflexible fitting. For the kuravirus SU10, our method explains most of the density in a semiautomated fashion. The results suggest rules for DNA turns in the end caps under which two discrete parameters determine the capsid inner diameter. We suggest that other kuraviruses viruses may follow the same winding scheme, producing a discrete rather than continuous spectrum of capsid inner diameters. Our software may be used to explain the published density maps of other double-stranded DNA viruses and uncover their genome packaging principles. Graphical Abstract

Most commercial laying hens suffer from sternum (keel) bone damage including deviations and fractures. X-raying hens, followed by segmenting and assessing the keel bone, is a key to automating the monitoring of keel bone condition. The aim of the current work is to train a deep learning model to segment the keel bone out of whole-body x-ray images. We obtained full-body x-ray images of laying hens (n = 1,051) and manually drew the outline of the keel bone on each image. Using the annotated images, a U-net model was then trained to segment the keel bone. The proposed model was evaluated using 5-fold cross validation. We obtained high segmentation accuracy (Dice coefficients of 0.88–0.90) repeatably over several validation folds. In conclusion, automatic segmentation of the keel bone from full-body x-ray images is possible with good accuracy. Segmentation is a requirement for automated measurements of keel geometry and density, which can subsequently be connected to susceptibility to keel deviations and fractures.

The interaction between human Growth Hormone (hGH) and hGH Receptor (hGHR) has basic relevance to cancer and growth disorders, and hGH is the scaffold for Pegvisomant, an anti-acromegaly therapeutic. For the latter reason, hGH has been extensively engineered by early workers to improve binding and other properties. We are particularly interested in E174 which belongs to the hGH zinc-binding triad; the substitution E174A is known to significantly increase binding, but to now no explanation has been offered. We generated this and several computationally-selected single-residue substitutions at the hGHR-binding site of hGH. We find that, while many successfully slow down dissociation of the hGH-hGHR complex once bound, they also slow down the association of hGH to hGHR. The E174A substitution induces a change in the Circular Dichroism spectrum that suggests the appearance of coiled-coiling. Here we show that E174A increases affinity of hGH against hGHR because the off-rate is slowed down more than the on-rate. For E174Y (and certain mutations at other sites) the slowdown in on-rate was greater than that of the off-rate, leading to decreased affinity. The results point to a link between structure, zinc binding, and hGHR-binding affinity in hGH.

Predicting the effect of mutations on protein-protein interactions is important for relating structure to function, as well as for in silico affinity maturation. The effect of mutations on protein-protein binding energy (ΔΔG) can be predicted by a variety of atomic simulation methods involving full or limited flexibility, and explicit or implicit solvent. Methods which consider only limited flexibility are naturally more economical, and many of them are quite accurate, however results are dependent on the atomic coordinate set used. In this work we perform a sequence and structure based search of the Protein Data Bank to find additional coordinate sets and repeat the calculation on each. The method increases precision and Positive Predictive Value, and decreases Root Mean Square Error, compared to using single structures. Given the ongoing growth of near-redundant structures in the Protein Data Bank, our method will only increase in applicability and accuracy.

In-frame decoding in the ribosome occurs through canonical or wobble Watson-Crick pairing of three mRNA codon bases (a triplet) with a triplet of anticodon bases in tRNA. Departures from the triplet-triplet interaction can result in frameshifting, meaning downstream mRNA codons are then read in a different register. There are many mechanisms to induce frameshifting, and most are insufficiently understood. One previously proposed mechanism is doublet decoding, in which only codon bases 1 and 2 are read by anticodon bases 34 and 35, which would lead to -1 frameshifting. In E. coli, tRNA(GCU)(Ser3) can induce -1 frameshifting at alanine (GCA) codons. The logic of the doublet decoding model is that the Ala codon's GC could pair with the tRNA(Ser3 ')s GC, leaving the third anticodon residue U36 making no interactions with mRNA. Under that model, a U36C mutation would still induce -1 frameshifting, but experiments refute this. We perform all-atom simulations of wild-type tRNA(Ser3), as well as a U36C mutant. Our simulations revealed a hydrogen bond between U36 of the anticodon and G1 of the codon. The U36C mutant cannot make this interaction, as it lacks the hydrogen-bond-donating H3. The simulation thus suggests a novel, non-doublet decoding mechanism for -1 frameshifting by tRNA(Ser3) at Ala codons.