Zika is a disease caused by the flavivirus Zika Virus (ZIKV), causing mostly mild symptoms such as fever, rash, headaches, joint and muscle pain and conjunctivitis. It has received renewed attention in the last years due to the 2015 outbreak in South America, which revealed its ability to also cause fetal malformation and Guillain-Barré Syndrome.
Most efforts against Zika and related infections focus on efforts to generate a vaccine or eradicate the urban vector, the mosquito Aedes aegypti, which is also responsible for transmission of other arboviruses such as Dengue, Yellow Fever and Chikungunya. However, once infected, there is no specific medicine available to combat the virus, and the options are restricted to symptomatic treatment.
One alternative to a long drug development program, which can take over 10 years with no guarantee of reaching a candidate, is to repurpose already available drugs to combat other diseases. In this webinar a recent effort will be shown in which Molecular Dynamics and Docking simulations were combined to scan a database of FDA-approved drugs for molecules that could potentially be used to fight Zika infection.
The search for bioactive compounds is the key step in early drug discovery. Among other techniques, the similarity principle (in the form of matched molecular pairs or free energy prediction), structure-based virtual screening, and of course experimental high throughput screening are applied. In this talk, our results related to the use of machine learning (ML) in these three design scenarios are summarized. How well does classical ML on matched molecular pairs affinity data perform? What signals do ML-based scoring functions for protein-ligand docking capture? How can we make use of ML in the evaluation of experimental screening data?
KNIME Analytics Platform is an open-source tool for data analysis, manipulation, visualization, and reporting. It integrates various life science extensions thus allowing one to deploy their cheminformatics and molecular modeling practices in a continuous flow. During the webinar we'll use an example to show how one can exploit these features to rapidly prototype ideas and share complex analyses with colleagues via KNIME.
We'll start by exploring the bioactivities of compounds versus five biological targets in order to find the molecules with a desired selectivity profile. We'll use a composite view with several interactive visualization elements in it. We'll look at the bioactivities using a parallel coordinates plot, a histogram, and a hierarchical cluster view and we'll display the individual molecules using tiles. We'll continue by explaining how one builds such a composite view in KNIME. Next, we'll pick the molecules with a desired selectivity profile and use them as queries to search Enamine REAL Space with the Search FTrees Fragment Space node of BioSolveIT extensions. In addition, we'll assess Tanimoto similarity of the obtained hits and calculate their physicochemical properties using the open source RDKit extensions. Last, but not the least we'll visually explore the results in order to prioritize the compounds.
Searching by similarity is a basic necessity in pharmaceutical research and myriad of different approaches are available. In this webinar we take a closer look at the FTrees method. It is based on a topological pharmacophore description of the molecule (the Feature Tree) and an algorithm that calculates the similarity based on an optimal mapping of two Feature Trees. We quickly introduce the science and technology at work behind the scenes and then focus on the discussion of two example applications in the context of drug discovery.
The first example will cover the results obtained in several retrospective virtual screenings (VS) carried out by applying FTrees and standard 2D fingerprint-based similarity calculations. The aim of this study was to assess the potential of these methods to prioritise active compounds. In a follow up study, we then tested a new option in FTrees, namely the ability to nominate more and less important features on a query molecule, on the basis of the same benchmark data.
In the second example, a comparable analysis was performed in a prospective VS of a challenging drug discovery project. While both methods unfortunately failed to identify new active compounds, there was very limited overlap in the compounds selected by FTrees and 2D-fingeprints, suggesting that FTrees and 2D fingerprints may be considered as complementary approaches to define compound similarity.