Data Science in Biology

Statistical methods allow training a computer to find meaningful patterns and correlations in enormously large volumes of biological data. The statistical results that passed laboratory verification can be taken into account for diagnostics, medical prescriptions or experiment planning.

The major problem of biological data processing is an elevated tendency of mathematical algorithms to overtraining (false positive predictions during production runs while having well trained models). There are two reasons for that - a very large dimension of the parameters space and, often, a limited training set. We are well aware of such effects and know how to minimize them.

Mathematical methods work only in connection with the understanding of biological mechanisms. It is crucial to correctly define the task. Our company has extensive biological expertise which, is absolutely required to analyze the biological data.

Our predictions go through a number of lab verifications and tuning cycles in order to achieve the highest level of veracity.