Motivation: Protein structure determination has primarily been performed using X-ray crystallography. To overcome the expensive cost, high attrition rate and series of trial-and-error settings, many in-silico methods have been developed to predict crystallization propensities of proteins based on their sequences. However, the majority of these methods build their predictors by extracting features from protein sequences, which is computationally expensive and can explode the feature space. We propose DeepCrystal, a deep learning framework for sequence-based protein crystallization prediction. It uses deep learning to identify proteins which can produce diffraction-quality crystals without the need to manually engineer additional biochemical and structural features from sequence. Our model is based on convolutional neural networks, which can exploit frequently occurring k-mers and sets of k-mers from the protein sequences to distinguish proteins that will result in diffraction-quality crystals from those that will not. Results: Our model surpasses previous sequence-based protein crystallization predictors in terms of recall, F-score, accuracy and Matthew’s correlation coefficient (MCC) on three independent test sets. DeepCrystal achieves an average improvement of 1.4, 12.1% in recall, when compared to its closest competitors, Crysalis II and Crysf, respectively. In addition, DeepCrystal attains an average improvement of 2.1, 6.0% for F-score, 1.9, 3.9% for accuracy and 3.8, 7.0% for MCC w.r.t. Crysalis II and Crysf on independent test sets. Availability and implementation: The standalone source code and models are available at https://github.com/elbasir/DeepCrystal and a web-server is also available at https://deeplearning-protein.qcri.org. Supplementary information: Supplementary data are available at Bioinformatics online.

The recent global health emergency caused by the coronavirus disease 2019 (COVID-19) pandemic has taken a heavy toll, both in terms of lives and economies. Vaccines against the disease have been developed, but the efficiency of vaccination campaigns worldwide has been variable due to challenges regarding production, logistics, distribution and vaccine hesitancy. Furthermore, vaccines are less effective against new variants of the SARS-CoV-2 virus and vaccination-induced immunity fades over time. These challenges and the vaccines’ ineffectiveness for the infected population necessitate improved treatment options, including the inhibition of the SARS-CoV-2 main protease (M^pro). Drug repurposing to achieve inhibition could provide an immediate solution for disease management. Here, we used structure-based virtual screening (SBVS) to identify natural products (from NP-lib) and FDA-approved drugs (from e-Drug3D-lib and Drugs-lib) which bind to the M^pro active site with high-affinity and therefore could be designated as potential inhibitors. We prioritized nine candidate inhibitors (e-Drug3D-lib: Ciclesonide, Losartan and Telmisartan; Drugs-lib: Flezelastine, Hesperidin and Niceverine; NP-lib: three natural products) and predicted their half maximum inhibitory concentration using DeepPurpose, a deep learning tool for drug–target interactions. Finally, we experimentally validated Losartan and two of the natural products as in vitro M^pro inhibitors, using a bioluminescence resonance energy transfer (BRET)-based M^pro sensor. Our study suggests that existing drugs and natural products could be explored for the treatment of COVID-19.

Sox9 is a fundamental sex-determining gene and the master regulator of chondrogenesis, and is involved in the development of various vital organs such as testes, kidney, heart and brain, and in skeletal development. Similar to other known Sox transcription factors, Sox9 recognizes and binds DNA with the consensus sequence C(T/A)TTG(T/A)(T/A) through the highly conserved HMG domain. Nonetheless, the molecular basis of the functional specificity of Sox9 in key developmental processes is still unclear. As an initial step towards a mechanistic understanding of Sox9 transcriptional regulation, the current work describes the details of the purification of the mouse Sox9 HMG domain (mSox9HMG), its crystallization in complex with a ChIP-Seq-identified FOXP2 promoter DNA element and the X-ray diffraction data analysis of this complex. The mSox9HMG–FOXP2 promoter DNA complex was crystallized by the hanging-drop vapour-diffusion method using 20% PEG 3350 in 200 mMsodium/potassium phosphate with 100 mMbis-tris propane at pH 8.5. The crystals diffracted to 2.7 Å resolution and the complex crystallized in the tetragonal space groupP4₁2₁2, with unit-cell parametersa=b= 99.49,c= 45.89 Å. Crystal-packing parameters revealed that asymmetric unit contained one mSox9HMG–FOXP2 promoter DNA complex with an estimated solvent content of 64%.