Neighbor-specific gene expression revealed from physically interacting cells during mouse embryonic development.

Development of multicellular organisms is orchestrated by persistent cell-cell communication between neighboring partners. Direct interaction between different cell types can induce molecular signals that dictate lineage specification and cell fate decisions. Current single-cell RNA-seq technology cannot adequately analyze cell-cell contact-dependent gene expression, mainly due to the loss of spatial information. To overcome this obstacle and resolve cell-cell contact-specific gene expression during embryogenesis, we performed RNA sequencing of physically interacting cells (PIC-seq) and assessed them alongside similar single-cell transcriptomes derived from developing mouse embryos between embryonic day (E) 7.5 and E9.5. Analysis of the PIC-seq data identified gene expression signatures that were dependent on the presence of specific neighboring cell types. Our computational predictions, validated experimentally, demonstrated that neural progenitor (NP) cells upregulate Lhx5 and Nkx2-1 genes, when exclusively interacting with definitive endoderm (DE) cells. Moreover, there was a reciprocal impact on the transcriptome of DE cells, as they tend to upregulate Rax and Gsc when in contact with NP cells. Using individual cell transcriptome data, we formulated a means of computationally predicting the impact of one cell type on the transcriptome of its neighboring cell types. We have further developed a distinctive spatial-t-distributed stochastic neighboring embedding to display the pseudospatial distribution of cells in a 2-dimensional space. In summary, we describe an innovative approach to study contact-specific gene regulation during embryogenesis.

DeepBindBC: A practical deep learning method for identifying native-like protein-ligand complexes in virtual screening.

Identifying native-like protein-ligand complexes (PLCs) from an abundance of docking decoys is critical for large-scale virtual drug screening in early-stage drug discovery lead searching efforts. Providing reliable prediction is still a challenge for most current affinity predicting models because of a lack of non-binding data during model training, lost critical physical-chemical features, and difficulties in learning abstract information with limited neural layers. In this work, we proposed a deep learning model, DeepBindBC, for classifying putative ligands as binding or non-binding. Our model incorporates information on non-binding interactions, making it more suitable for real applications. ResNet model architecture and more detailed atom type representation guarantee implicit features can be learned more accurately. Here, we show that DeepBindBC outperforms Autodock Vina, Pafnucy, and DLSCORE for three DUD.E testing sets. Moreover, DeepBindBC identified a novel human pancreatic α-amylase binder validated by a fluorescence spectral experiment (Ka = 1.0 × 105 M). Furthermore, DeepBindBC can be used as a core component of a hybrid virtual screening pipeline that incorporating many other complementary methods, such as DFCNN, Autodock Vina docking, and pocket molecular dynamics simulation. Additionally, an online web server based on the model is available at http://cbblab.siat.ac.cn/DeepBindBC/index.php for the user's convenience. Our model and the web server provide alternative tools in the early steps of drug discovery by providing accurate identification of native-like PLCs.

Validation of Deep Learning-Based DFCNN in Extremely Large-Scale Virtual Screening and Application in Trypsin I Protease Inhibitor Discovery.

Computational methods with affordable computational resources are highly desirable for identifying active drug leads from millions of compounds. This requires a model that is both highly efficient and relatively accurate, which cannot be achieved by most of the current methods. In real virtual screening (VS) application scenarios, the desired method should perform much better in selecting active compounds by prediction than by random chance. Here, we systematically evaluate the performance of our previously developed DFCNN model in large-scale virtual screening, and the results show our method has approximately 22 times the success rate compared to the random chance on average with a score cutoff of 0.99. Of the 102 test cases, 10 cases have more than 98 times the success rate of a random guess. Interestingly, in three cases, the prediction success rate is 99 times that of a random guess by a score cutoff of 0.99. This indicates that in most situations after our extremely large-scale VS, the dataset can be reduced 20 to 100 times for the next step of virtual screening based on docking or MD simulation. Furthermore, we have employed an experimental method to verify our computational method by finding several activity inhibitors for Trypsin I Protease. In addition, we also show its proof-of-concept application in de novo drug screening. The results indicate the massive potential of this method in the first step of the real drug development workflow. Moreover, DFCNN only takes about 0.0000225s for one protein-compound prediction on average with 80 Intel CPU cores (2.00 GHz) and 60 GB RAM, which is at least tens of thousands of times faster than AutoDock Vina or Schrödinger high-throughput virtual screening. Additionally, an online webserver based on DFCNN for large-scale screening is available at http://cbblab.siat.ac.cn/DFCNN/index.php for the convenience of the users.

Flower lose, a cell fitness marker, predicts COVID-19 prognosis.

Risk stratification of COVID-19 patients is essential for pandemic management. Changes in the cell fitness marker, hFwe-Lose, can precede the host immune response to infection, potentially making such a biomarker an earlier triage tool. Here, we evaluate whether hFwe-Lose gene expression can outperform conventional methods in predicting outcomes (e.g., death and hospitalization) in COVID-19 patients. We performed a post-mortem examination of infected lung tissue in deceased COVID-19 patients to determine hFwe-Lose's biological role in acute lung injury. We then performed an observational study (n = 283) to evaluate whether hFwe-Lose expression (in nasopharyngeal samples) could accurately predict hospitalization or death in COVID-19 patients. In COVID-19 patients with acute lung injury, hFwe-Lose is highly expressed in the lower respiratory tract and is co-localized to areas of cell death. In patients presenting in the early phase of COVID-19 illness, hFwe-Lose expression accurately predicts subsequent hospitalization or death with positive predictive values of 87.8-100% and a negative predictive value of 64.1-93.2%. hFwe-Lose outperforms conventional inflammatory biomarkers and patient age and comorbidities, with an area under the receiver operating characteristic curve (AUROC) 0.93-0.97 in predicting hospitalization/death. Specifically, this is significantly higher than the prognostic value of combining biomarkers (serum ferritin, D-dimer, C-reactive protein, and neutrophil-lymphocyte ratio), patient age and comorbidities (AUROC of 0.67-0.92). The cell fitness marker, hFwe-Lose, accurately predicts outcomes in COVID-19 patients. This finding demonstrates how tissue fitness pathways dictate the response to infection and disease and their utility in managing the current COVID-19 pandemic.

CRISPR-assisted detection of RNA-protein interactions in living cells.

We have developed CRISPR-assisted RNA-protein interaction detection method (CARPID), which leverages CRISPR-CasRx-based RNA targeting and proximity labeling to identify binding proteins of specific long non-coding RNAs (lncRNAs) in the native cellular context. We applied CARPID to the nuclear lncRNA XIST, and it captured a list of known interacting proteins and multiple previously uncharacterized binding proteins. We generalized CARPID to explore binders of the lncRNAs DANCR and MALAT1, revealing the method's wide applicability in identifying RNA-binding proteins.

DeepBindPoc: a deep learning method to rank ligand binding pockets using molecular vector representation.

Accurate identification of ligand-binding pockets in a protein is important for structure-based drug design. In recent years, several deep learning models were developed to learn important physical-chemical and spatial information to predict ligand-binding pockets in a protein. However, ranking the native ligand binding pockets from a pool of predicted pockets is still a hard task for computational molecular biologists using a single web-based tool. Hence, we believe, by using closer to real application data set as training and by providing ligand information, an enhanced model to identify accurate pockets can be obtained. In this article, we propose a new deep learning method called DeepBindPoc for identifying and ranking ligand-binding pockets in proteins. The model is built by using information about the binding pocket and associated ligand. We take advantage of the mol2vec tool to represent both the given ligand and pocket as vectors to construct a densely fully connected layer model. During the training, important features for pocket-ligand binding are automatically extracted and high-level information is preserved appropriately. DeepBindPoc demonstrated a strong complementary advantage for the detection of native-like pockets when combined with traditional popular methods, such as fpocket and P2Rank. The proposed method is extensively tested and validated with standard procedures on multiple datasets, including a dataset with G-protein Coupled receptors. The systematic testing and validation of our method suggest that DeepBindPoc is a valuable tool to rank near-native pockets for theoretically modeled protein with unknown experimental active site but have known ligand. The DeepBindPoc model described in this article is available at GitHub (https://github.com/haiping1010/DeepBindPoc) and the webserver is available at (http://cbblab.siat.ac.cn/DeepBindPoc/index.php).

DeepBindRG: a deep learning based method for estimating effective protein-ligand affinity.

Proteins interact with small molecules to modulate several important cellular functions. Many acute diseases were cured by small molecule binding in the active site of protein either by inhibition or activation. Currently, there are several docking programs to estimate the binding position and the binding orientation of protein-ligand complex. Many scoring functions were developed to estimate the binding strength and predict the effective protein-ligand binding. While the accuracy of current scoring function is limited by several aspects, the solvent effect, entropy effect, and multibody effect are largely ignored in traditional machine learning methods. In this paper, we proposed a new deep neural network-based model named DeepBindRG to predict the binding affinity of protein-ligand complex, which learns all the effects, binding mode, and specificity implicitly by learning protein-ligand interface contact information from a large protein-ligand dataset. During the initial data processing step, the critical interface information was preserved to make sure the input is suitable for the proposed deep learning model. While validating our model on three independent datasets, DeepBindRG achieves root mean squared error (RMSE) value of pKa (-logKd or -logKi) about 1.6-1.8 and R value around 0.5-0.6, which is better than the autodock vina whose RMSE value is about 2.2-2.4 and R value is 0.42-0.57. We also explored the detailed reasons for the performance of DeepBindRG, especially for several failed cases by vina. Furthermore, DeepBindRG performed better for four challenging datasets from DUD.E database with no experimental protein-ligand complexes. The better performance of DeepBindRG than autodock vina in predicting protein-ligand binding affinity indicates that deep learning approach can greatly help with the drug discovery process. We also compare the performance of DeepBindRG with a 4D based deep learning method "pafnucy", the advantage and limitation of both methods have provided clues for improving the deep learning based protein-ligand prediction model in the future.

IVS2vec: A tool of Inverse Virtual Screening based on word2vec and deep learning techniques.

Inverse Virtual Screening is a powerful technique in the early stage of drug discovery process. This technique can provide important clues for biologically active molecules, which is useful in the following researches of durg discovery. In this work, combining with Word2vec, a natural language processing technique, dense fully connected neural network (DFCNN) algorithm is utilized to build up a prediction model. This model is able to perform a binary classification. Based on the query molecule, the input protein candidates can be classified into two subsets. One set is that potential targets with high possibilities to bind with the query molecule and the other one is that the proteins with low possibilities to bind with the query molecule. This model is named as IVS2vec. IVS2vec also can output a score reflecting binding possibility of the association between a protein and a molecule, which is useful to improve efficiency of research. We applied IVS2vec on several databases related to drug development and shown that our model can detect possible therapeutic targets. In addition, our model can identify targets related to adverse drug reactions which is useful to improve medication safety and repurpose drugs. Moreover, IVS2vec can give a very fast speed to perform prediction jobs. It is suitable for processing a large number of compounds in the chemical databases. We also find that IVS2vec has potential capabilities and outperform other state-of-the-art docking tools such as Autodock vina. In this study, IVS2vec brings many convincing results than Autodock vina in the reverse target searching case of Quercetin.