Inflammation is a protective biological response to body/tissue damage that involves immune cells, blood vessels and molecular mediators. In this work, we constructed the pathway network of inflammation, including 11 sub-pathways of inflammatory factors. Pathway-based network efficiency and network flux were adopted to evaluate drug efficacy. By using approved and experimentally validated anti-inflammatory drugs as training sets, a predictive model was built to screen potential anti-inflammatory drugs from approved drugs in DrugBank. This drug repositioning approach would bring a fast and cheap way to find new indications for approved drugs. Moreover, molecular phenomics profiles of the expression of inflammatory factors will provide new insight into the drug mechanism of action.

Cancer is a complex disease, known medically as malignant neoplasm. Natural products (NPs) play a very important role in anticancer drug discovery and a large number of NPs have been proven to have potential anticancer effects. Compared with newly synthesized chemical compounds, NPs show a favorable profile in terms of their absorption and metabolism in the body with low toxicity. Searching for multi-target natural drugs can be regarded as a solution to improve therapeutic efficacy and safety. In this work, we collected 104 cancer-associated target proteins from the Protein Data Bank. Based on the Universal Natural Products Database, all of the NPs were docked to 104 cancer-associated target proteins. Then we explored the potential of NPs and several herbs in anticancer drug discovery by using a network-based multi-target computational approach. The NPs with the most potential for anticancer drug discovery and their indications were predicted based on a docking score-weighted prediction model. We also explored the interactions between NPs and cancer target proteins to find the pathological networks, potential drug candidates and new indications.

Chemogenomics focuses on the interactions between biologically active molecules and protein targets for drug discovery. Carbohydrates are the most abundant compounds in natural products. Compared with other drugs, the carbohydrate drugs show weaker side effects. Searching for multi-target carbohydrate drugs can be regarded as a solution to improve therapeutic efficacy and safety. In this work, we collected 60344 carbohydrates from the Universal Natural Products Database (UNPD) and explored the chemical space of carbohydrates by principal component analysis. We found that there is a large quantity of potential lead compounds among carbohydrates. Then we explored the potential of carbohydrates in drug discovery by using a network-based multi-target computational approach. All carbohydrates were docked to 2389 target proteins. The most potential carbohydrates for drug discovery and their indications were predicted based on a docking score-weighted prediction model. We also explored the interactions between carbohydrates and target proteins to find the pathological networks, potential drug candidates and new indications.

Fluoroquinolones play an important role in the treatment of serious bacterial infections, but at the same time they could lead to cardiac toxicity due to the blockage of the HERG potassium channel, which even leads to the withdrawal of some fluoroquinolones. Blockage of the HERG potassium channel by drugs or drug-like compounds has become a critical problem in drug discovery. Though there were large amounts of bioactivity data of fluoroquinolones on the blockage of HERG, little structural basis of binding of blockers to the HERG channel was known. Here, we combined molecular docking, molecular dynamics simulations, free energy calculations and binding energy decomposition analysis to explore the binding modes of fluoroquinolones in the HERG potassium channel. The calculated binding free energies were consistent with the experimental binding affinities. Our results showed that the CH3 group in MX was favorable for the binding to the HERG channel, while Tyr652 and Phe656 were critical for the hydrophobic interaction between fluoroquinolones and the HERG channel. We expected that our results of calculation could provide important insights for the rational design and discovery of drugs.

In this work, we prepared a monolithic and surface initiated molecularly imprinted polymeric (MIP) column for HPLC and explored its application for template separation from plant extract. The silica beads (40–60 μm) were coupled with initiator on the surface and then packed in to a stainless steel HPLC column. The pre-polymerization mixture (the template, functional monomer and crosslinker were emodin, acrylamide and divinylbenzene, respectively) was injected into the column and polymerized by thermal initiation. The prepared MIP column exhibited excellent retention capability and large imprinted factor for template (the retention time and imprinted factor for emodin on MIP column were 16.26 min and 7.21, respectively). Moreover, the emodin-molecularly imprinted polymeric column could be applied to separate emodin from alcohol extract of Rheum palmatum L. at semi-preparative scale and 1.2 mg of emodin was obtained in 4 h.Highlights► A monolithic and surface initiated molecularly imprinted polymeric column for HPLC was prepared by in situ synthesis. ► The emodin-MIP column exhibited excellent retention capability and large imprinted factor for its template. ► The emodin-MIP column could be applied to separate emodin from alcohol extract of Rheum palmatum L. at semi-preparative scale. ► This method used a small quantity of template, which would be important for some natural products if they are difficult to obtain.