POSTER ABSTRACT / DETAILS:
Pharmacological reactivation of the γ-globin gene for the production of fetal haemoglobin (HbF) is a very promising therapeutic avenue for β-thalassaemia. Increased production of γ-globin can ameliorate the symptoms of the disease by partly substituting the non-functional β-globin gene, and restoring the balance between α and non-α globin chains. Drugs currently available for this purpose have limited application due to moderate therapeutic properties, variable patient response and potential cytotoxic effects. The main aim of this work is to discover agents which have higher efficacy and are safer than existing drugs through the development of a novel pathway for drug discovery and design involving the use of molecular modeling and data mining.
Towards this goal we have for long been involved in the in silico exploration of various molecular patterns for the identification of novel HbF inducers.
For this purpose we have combined data mining, machine learning with similarity search and virtual screening techniques to understand the structural characteristics that affect the reactivation of HbF from different molecular patterns. We have developed KNIME workflows (analyzing data and automatically reporting) to support customized needs of this work.
We used existing and developed new custom-made KNIME nodes to develop a customized processing, analysis, and exploration platform. A model has been built using in vitro screening data in K562 human erythroleukaemia cultures from different molecular structures in order to study their ability to activate the γ-globin gene. K562 human erythroleukaemic cell line has the potential to highly express the embryo-fetal globin genes such as ζ, ε, and γ-globin genes. This characteristic makes K562 cells a useful model cell line for the study of compounds that are potential γ-globin inducers.
The produced model was subsequently tested for accurate class predictions with an independent set of samples, and agreement of results was assessed.
This work is supported by funding under the Seven Research Framework Programme of the European Union. Project THALAMOSS (HEALTH.2012.1.2-1 Grant agreement no: 306201)