A fast machine learning dataloader for epigenetic tracks from BigWig files.

SUMMARY: We created bigwig-loader, a data-loader for epigenetic profiles from BigWig files that decompresses and processes information for multiple intervals from multiple BigWig files in parallel. This is an access pattern needed to create training batches for typical machine learning models on epigenetics data. Using a new codec, the decompression can be done on a graphical processing unit (GPU) making it fast enough to create the training batches during training, mitigating the need for saving preprocessed training examples to disk. AVAILABILITY AND IMPLEMENTATION: The bigwig-loader installation instructions and source code can be accessed at https://github.com/pfizer-opensource/bigwig-loader.

NMR structure and dynamics of Q4DY78, a conserved kinetoplasid-specific protein from Trypanosoma cruzi.

The 106-residue protein Q4DY78 (UniProt accession number) from Trypanosoma cruzi is highly conserved in the related kinetoplastid pathogens Trypanosoma brucei and Leishmania major. Given the essentiality of its orthologue in T. brucei, the high sequence conservation with other trypanosomatid proteins, and the low sequence similarity with mammalian proteins, Q4DY78 is an attractive protein for structural characterization. Here, we solved the structure of Q4DY78 by solution NMR and evaluated its backbone dynamics. Q4DY78 is composed of five α -helices and a small, two-stranded antiparallel ß-sheet. The backbone RMSD is 0.22 ± 0.05 Å for the representative ensemble of the 20 lowest-energy structures. Q4DY78 is overall rigid, except for N-terminal residues (V8 to I10), residues at loop 4 (K57 to G65) and residues at the C-terminus (F89 to F112). Q4DY78 has a short motif FPCAP that could potentially mediate interactions with the host cytoskeleton via interaction with EVH1 (Drosophila Enabled (Ena)/Vasodilator-stimulated phosphoprotein (VASP) homology 1) domains. Albeit Q4DY78 lacks calcium-binding motifs, its fold resembles that of eukaryotic calcium-binding proteins such as calcitracin, calmodulin, and polcacin Bet V4. We characterized this novel protein with a calcium binding fold without the capacity to bind calcium.

grünifai: interactive multiparameter optimization of molecules in a continuous vector space.

SUMMARY: Optimizing small molecules in a drug discovery project is a notoriously difficult task as multiple molecular properties have to be considered and balanced at the same time. In this work, we present our novel interactive in silico compound optimization platform termed grünifai to support the ideation of the next generation of compounds under the constraints of a multiparameter objective. grünifai integrates adjustable in silico models, a continuous representation of the chemical space, a scalable particle swarm optimization algorithm and the possibility to actively steer the compound optimization through providing feedback on generated intermediate structures. AVAILABILITY AND IMPLEMENTATION: Source code and documentation are freely available under an MIT license and are openly available on GitHub (https://github.com/jrwnter/gruenifai). The backend, including the optimization method and distribution on multiple GPU nodes is written in Python 3. The frontend is written in ReactJS.

Structure of outer membrane protein G in lipid bilayers.

ß-barrel proteins mediate nutrient uptake in bacteria and serve vital functions in cell signaling and adhesion. For the 14-strand outer membrane protein G of Escherichia coli, opening and closing is pH-dependent. Different roles of the extracellular loops in this process were proposed, and X-ray and solution NMR studies were divergent. Here, we report the structure of outer membrane protein G investigated in bilayers of E. coli lipid extracts by magic-angle-spinning NMR. In total, 1847 inter-residue 1H-1H and 13C-13C distance restraints, 256 torsion angles, but no hydrogen bond restraints are used to calculate the structure. The length of ß-strands is found to vary beyond the membrane boundary, with strands 6-8 being the longest and the extracellular loops 3 and 4 well ordered. The site of barrel closure at strands 1 and 14 is more disordered than most remaining strands, with the flexibility decreasing toward loops 3 and 4. Loop 4 presents a well-defined helix.

Alterations in creatine metabolism observed in experimental autoimmune myocarditis using ex vivo proton magic angle spinning MRS.

Experimental autoimmune myocarditis (EAM) in rodents is an accepted model of myocarditis and dilated cardiomyopathy (DCM). Altered metabolism is thought to play an important role in the pathogenesis of DCM and heart failure (HF). Study of the metabolism may provide new diagnostic information and insights into the mechanisms of myocarditis and HF. Proton MRS ((1)H-MRS) has not yet been used to study the changes occurring in myocarditis and subsequent HF. We aimed to explore the changes in creatine metabolism using this model and compare them with the findings in healthy animals. Myocardial function of male young Lewis rats with EAM was quantified by performing left ventricular ejection fraction (LVEF) analysis in short-axis cine images throughout the whole heart. Inflammatory cellular infiltrate was assessed by immunohistochemistry. Myocardial tissue was analyzed using ex vivo proton magic angle spinning MRS ((1)H-MAS-MRS). Myocarditis was confirmed histologically by the presence of an inflammatory cellular infiltrate and CD68 positive staining. A significant increase in the metabolic ratio of Tau/tCr (taurine/total creatine) obtained by (1)H-MAS-MRS was observed in myocarditis compared with healthy controls (21 d acute EAM, 4.38 (±0.23); 21 d control, 2.84 (±0.08); 35 d chronic EAM, 4.47 (±0.83); 35 d control, 2.59 (±0.38); P < 0.001). LVEF was reduced in diseased animals (EAM, 55.2% (±11.3%); control, 72.6% (±3.8%); P < 0.01) and correlated with Tau/tCr ratio (R = 0.937, P < 0.001). Metabolic alterations occur acutely with the development of myocarditis. Myocardial Tau/tCr ratio as detected by (1)H-MRS correlates with LVEF and is able to differentiate between healthy myocardium and myocardium from rats with EAM.

Rapid proton-detected NMR assignment for proteins with fast magic angle spinning.

Using a set of six (1)H-detected triple-resonance NMR experiments, we establish a method for sequence-specific backbone resonance assignment of magic angle spinning (MAS) nuclear magnetic resonance (NMR) spectra of 5-30 kDa proteins. The approach relies on perdeuteration, amide (2)H/(1)H exchange, high magnetic fields, and high-spinning frequencies (ωr/2π ≥ 60 kHz) and yields high-quality NMR data, enabling the use of automated analysis. The method is validated with five examples of proteins in different condensed states, including two microcrystalline proteins, a sedimented virus capsid, and two membrane-embedded systems. In comparison to contemporary (13)C/(15)N-based methods, this approach facilitates and accelerates the MAS NMR assignment process, shortening the spectral acquisition times and enabling the use of unsupervised state-of-the-art computational data analysis protocols originally developed for solution NMR.

Out-and-back 13C-13C scalar transfers in protein resonance assignment by proton-detected solid-state NMR under ultra-fast MAS.

We present here (1)H-detected triple-resonance H/N/C experiments that incorporate CO-CA and CA-CB out-and-back scalar-transfer blocks optimized for robust resonance assignment in biosolids under ultra-fast magic-angle spinning (MAS). The first experiment, (H)(CO)CA(CO)NH, yields (1)H-detected inter-residue correlations, in which we record the chemical shifts of the CA spins in the first indirect dimension while during the scalar-transfer delays the coherences are present only on the longer-lived CO spins. The second experiment, (H)(CA)CB(CA)NH, correlates the side-chain CB chemical shifts with the NH of the same residue. These high sensitivity experiments are demonstrated on both fully-protonated and 100%-H(N) back-protonated perdeuterated microcrystalline samples of Acinetobacter phage 205 (AP205) capsids at 60 kHz MAS.

Three-dimensional deuterium-carbon correlation experiments for high-resolution solid-state MAS NMR spectroscopy of large proteins.

Well-resolved (2)H-(13)C correlation spectra, reminiscent of (1)H-(13)C correlations, are obtained for perdeuterated ubiquitin and for perdeuterated outer-membrane protein G (OmpG) from E. coli by exploiting the favorable lifetime of (2)H double-quantum (DQ) states. Sufficient signal-to-noise was achieved due to the short deuterium T (1), allowing for high repetition rates and enabling 3D experiments with a (2)H-(13)C transfer step in a reasonable time. Well-resolved 3D (2)H(DQ)-(13)C-(13)C correlations of ubiquitin and OmpG were recorded within 3.5 days each. An essentially complete assignment of (2)H(DQα) shifts and of a substantial fraction of (2)H(DQß) shifts were obtained for ubiquitin. In the case of OmpG, (2)H(DQα) and (2)H(DQß) chemical shifts of a considerable number of threonine, serine and leucine residues were assigned. This approach provides the basis for a general heteronuclear 3D MAS NMR assignment concept utilizing pulse sequences with (2)H(DQ)-(13)C transfer steps and evolution of deuterium double-quantum chemical shifts.