Molecular Modeling Services
molecular modeling services
Scientific Software Development Services
Scientific Software Development Services
Corporate Alliances & Partnerships
Corporate Alliances & Partnerships
Services

Over the last 10 years, Intelligent Pharma has acquired extensive experience in scientific software and solutions development specific to the needs that arise during research & development. As a life sciences company ourselves, we understand first-hand the data and technologies our clients use and the challenges they face.

We work with technologies that intensively use computational resources. Some examples are virtual screening of molecules, molecular dynamics, and Deep Neural Networks (DNNs). Our team has extensive experience in the use of supercomputing (HPC) by using our own in house supercomputer or the cloud (or a combination of both).

With communication and flexibility as core values, and using Agile methodologies, our team can create solutions specific to the needs of life sciences companies. Furthermore, our IT team is composed out of engineers and doctors in Computer Science, guaranteeing a high quality, professional, maintainable and scalable software. Your project will also benefit from a close collaboration between our IT team and the other departments present at Intelligent Pharma (chemists, biologists…).

The research process involves the generation of a large amount of heterogeneous data. Intelligent Pharma has multiple tools at its disposal in order to delve into this data, organize it, and create specific solutions for its unification, management, analysis, and subsequent knowledge extraction.

Big data is an extremely powerful tool that allows you to perform simulations and predictions in very diverse areas. We can take your data sets and objectives and design and implement a data mining strategy using cutting edge methodologies and tools. Talk to us!

EXAMPLES:
Some examples of our work include an intelligent search solution for managing preclinical oncology model databases, analogous identification in a petrochemical database and a wide range of solutions for managing and analyzing chemical compounds. These are described in more detail below:
Oncology models portfolio management tool

Our client required a very intuitive web application that would allow customers to easily find and compare data (gene expression, mutations, tumor characteristics, histology images and standard of care data) of preclinical murine oncology models and display this information in a simple and interactive way, to facilitate the selection of the most appropriate model for the customer’s research.

The main challenge of this project was that the model information was located in heterogeneous data sources, and the client required a single point of entry into these data sources.

The solution that we created was able to abstract the model information from the different data sources, index them, and implement a complex search engine among the data sources to compare and analyze them using a simple graphical interface. The result was a simple google-like interface that allows customers to execute simple searches in order to locate specific and relevant model information, regardless of the complexity of the parent databases.

Furthermore, the solution included a business intelligence module that tracked user search and usage information (such as search frequency, searches without results) to help drive the business strategy, understand better the customers´ needs and allow sales representatives to offer training and support.

Petrochemical reservoir databases management tool

The goal of this project was to unify a large number of heterogeneous petrochemical reservoir databases into a single database which could detect analogous reservoirs among the different databases, manage upgrades and eventual changes in the data model, perform statistical analyses of the data, and generate reports.

This project also included unit detection and standardization, synonym management, configuration of custom validation functions for the data, and users and roles management.

The solution was a usable and intuitive web application over a relational database.

Chemical compounds management tool

The aim of this project was to create a simple compound management tool which could compile results from experiments over a number of compounds developed by different companies in an international collaboration research project.

The tool was a web application over a relational database that allowed the creation, modification, and deletion of compounds, while also being able to perform physical property calculations and statistical analysis. The tool also provided the functionality of managing users, roles, and permissions over the view of compounds and the operations therein.

One characteristic that distinguishes our computational services from other providers is our foundation consisting of internally developed solutions. Our IT team at Intelligent Pharma uses all its artificial intelligence and statistical skills to collaborate with scientists and develop state-of-the-art scientific software, especially in the field of computational chemistry.

Our internal solutions include virtual screening, QSAR, hit to lead technology based on genetic algorithms, and a polypharmacology prediction tool.

EXAMPLES:
More detail about our flagship software solutions can be found here.
Prometheus

Prometheus is a tool that combines the creation of de novo compound libraries with artificial learning and evolutionary computation to explore new chemical spaces and create new and revolutionary drug candidates. Prometheus takes into account different parameters when optimizing compounds, such as their synthesizability or drug-likeness, and also has the flexibility to use other more specific evaluation functions, such as a molecules activity against a particular target.

GIMIF

GIMIF (Grid Independent Molecular Interaction Fields) are three dimensional descriptors which allow a 3D-structure-quantitative analysis of compounds. 3D-QSAR models based on GIMIF are more interpretable and efficient, as no preliminary alignment of compounds is required and direct correlation between functional groups and descriptors is available. IP validated these 3D descriptors for its own models and performed a hepatoxicity model of CYP450, with a R2= 0.78 and a Q2=0.81.

Nowadays, researchers have vast amounts of data at their disposal. This data comes from both freely available sources and privately owned databases. Moreover, this data can contain key insight into the drug discovery process that might not be easily assessed by mere observation. Machine learning techniques present themselves as the ideal tools for this kind of task, being able to extract the underlying structure of data. This knowledge can be then applied to prediction tasks.

Our expertise includes support vector machines, PLS, deep learning, and random forests for tasks such as QSAR modelling, prediction of the mechanism of action of molecules, and predictions of drug-drug interactions.

EXAMPLES:
Medea

This technology allows us to build QSAR models from a variety of descriptors and mathematical methods. The descriptors supported by the technology are rCDK, ECFP and GRINDs, among others. Support vector machines (both regression and classification) and partial least squares methods are included in this technology.

Mechanism of Action (MoA)

Our technology PYTHIA, can predict the mechanism of action of chemical compounds thanks to a model based on Deep Learning. The model is trained on millions of activities registered between hundreds of thousands compounds and thousands of targets. This model allows us to predict the activity (or lack of) of a compound against around two thousand targets.

Drug-Drug Interactions (DDIs)

Currently in the research phase, this new technology will allow the prediction of drug-drug interactions between a drug candidate and already approved drugs. To do so, we are training a random forest model with drug-protein interaction data (from sources as ChEMBL, PubChem, etc), as well as metabolic pathway data extracted from the Reactome database.

With core values consisting of effective communication and flexibility, and an attitude towards an agile methodology of development, Intelligent Pharma is able to create solutions specific for the needs of the different stages of the research process, offering usable and professional technologies.

By a thorough analysis of your needs, we propose a software solution and interface design that fits best to your needs. Our solutions are reusable and maintainable. We work together with you during the development process to guaranty a solution as you envisioned. Once finalized the solution is deployed over the internet or installed in your internal infrastructure, ready to use.