Funding: European Regional Development Fund (ERDF) “On Implementation of Activity 1.1.1.2 “Post-doctoral Research Aid” of the Specific Aid Objective 1.1.1 “To increase the research and innovative capacity of scientific institutions of Latvia and the ability to attract external financing, investing in human resources and infrastructure” of the Operational Programme “Growth and Employment”
Project Title: “Identification of clinical subgroups of Type 2 diabetes mellitus and application of pharmacogenetics in development of personalized antidiabetic therapy”
Project No. 1.1.1.2/VIAA/2/18/287
Period: 1st March 2019 – 31 October 2022
Project costs: 133 806,00 EUR
Project implementer: Dr. biol. Raitis Pečulis
Project Summary:
The project proposes subtyping of Type 2 Diabetes mellitus (T2DM) patients and evaluation of effects of pharmacogenetic markers on patient-significant therapy outcomes and prevention of complications. The stratification of T2DM in a sample of ~ 3 800 patients will be assessed by using data-driven cluster analysis; pharmacogenetic research, investigating the association between clinical outcomes and T2DM susceptibility scores, will be investigated in the subgroup of ~ 1 000 patients with GWAS data. The acquired dataset will be analysed by applying classical statistical analysis and supervised machine learning methods (decision trees and simple neural networks). Identified significant clinical variables and genetic factors will be used to create a model for prediction of the treatment outcomes. Replication of findings will be performed in the GoDARTS dataset with 10 000 GWAS available. Project results will potentially facilitate the development of patient-centred treatment and identification of groups in need for early intensive therapy.
Information published 01.03.2019.
Progress of the project:
1 March 2019 – 31 May 2019
During the 01.03.2019.-31.05.2019. the spectrum of variables and metadata from ‘‘Register of patients with particular diseases, including patients with diabetes mellitus’’ and Genome Database of Latvian Population (LGDB) as well as from Death causes database of the Centre for Disease Prevention and Control were assessed and data analysis and storage methods chosen. The application for the approval of the project by The Central Medical Ethics Committee of Latvia was submitted. The preliminary application for additional retrieval data from UK Biobank data was prepared. The draft of a publication entitled “All-cause mortality risk in adults with and without type 2 diabetes: findings from the Genome Database of the Latvian Population” was prepared according to guidelines of BMC Public Health journal.
Information published 31.05.2019.
Progress of the project:
1 June 2019 – 31 August 2019
Database with relevant publications was created in reference program Mendeley and organised for research purposes. The clinical phenotypes and anthropometric measurements were investigated regarding susceptibility to complications and response to therapy. Possible associations were analysed in public databases The NHGRI-EBI GWAS Catalog and GWAS Central. Data were investigated regarding methods used in previous studies. R packages like ggplot2, dyplr, foreign, cluster and others were investigated for the purposes of the analysis of the data and visualization of the results.
Information published 30.08.2019.
Progress of the project:
1 September 2019 – 30 November 2019
In September the statistical analysis were finished and preparation for mobility started. Additional time was dedicated for learning to work with Python and LINUX. From 01.10.2019.-30.11.2019 resercher spent in Estonian Genome Center. During stay work started to learn analyse GWAs and work with data analysis within cluster from University of Tartu. Data analysis and reserch methods were compared regarding efficiency and quality.
Information published 29.11.2019.
Progress of the project:
1 December 2019 – 29 February 2020
From 01.12.2019.-29.02.2019 researcher spent in Estonian Genome Center. The data comparison was finished and list of phenotypes was revisited and additional tables and pictures were created. Additional time was dedicated to learn R and GWAs. The comparison revealed important phenotypes affecting development and progression of diabetes in the Latvian cohorts (T1DM and T2DM), as well as the methods learned in Estonia were applied for the connection of the datasets.
Information published 28.02.2020.
Progress of the project:
1 May 2020 – 31 July 2020
After the replacement of project implementer the compilation of all currently obtained data was performed for type 2 diabetes patients. It was followed by GWAS analysis and creation of a polygenic risk score model for patients of Latvian population as well as creation of a model of population stratification of Latvia which could aid in further GWAS analyses in Latvian population. Compilation of currently available phenotype and genotype data of type 2 diabetes patients was commenced with the aim to analyse data in subgroups of patients which will be based on type 2 diabetes late complications. ESHG 2020 e-conference was attended by the current project implementer.
Information published 31.07.2020.
Progress of the project:
1 August 2020 – 31 October 2020
Genotyping of additional diabetes patients were performed using the Human GSA genotyping chip. Association analysis of type 2 diabetes patients with vascular complications and their genotypes was performed. Analysis in distinct subgroups of complications shows variability of genetic background. Obtained data have been prepared for publication and graphical material created. Possibilities to study and use neuronal network analysis in understanding role of genetics markers in diabetes complications are investigated.
Information published 30.10.2020.
Progress of the project:
1 November 2020 – 31 January 2021
Publication about diabetes complicatons was accepted and has been published: „Novel susceptibility loci identified in a genome-wide association study of type 2 diabetes complications in population of Latvia”, Ustinova, M., Peculis, R., Rescenko, R., Pirags, V., Klovins, J. BMC Medical Genomics, 2021, 14(1), 18
Phenotype and pharmacogenomic data of available sulfanilurea therapy patients were compiled for international colaboration study. Analysis of latest whole genome genotyping data was commenced. Midterm results of the project were evaluated by the independent international expert and positive marks were acquired.
Information published 29.01.2021.
Progress of the project:
1 February 2021 – 30 April 2021
The results of the project have been accepted for presenting in an international conference ESHG 2021. After publication of diabetes complication data, the ongoing work to summarizing and systematization of farmakogenetics data for analysis was continued. Latest literature on farmacogenetics of diabetes, metabolic syndrome and obesity was studied and compiled. Also, studies of using artificial neural networks for developing a farmacogenetics model of diabetes were continued. For application of artificial neural networks software Matlab Neural Networks Toolbox and Neuroph (Java) was evaluated.
Information published 30.04.2021.
Progress of the project:
1 May 2021 – 31 July 2021
Additional cohort of novel patients has been selected for genotyping in farmacogenetics study. Software for neuronal networks has been studied and tested. Preparation of data for model construction is commencing. Possibility of publishing review article about “Pharmacogenetics, clustering and complications” is being investigated.
Information published 30.07.2021.
Progress of the project:
1 August 2021 – 31 October 2021
During this project period, additional genotyping of patients with type 2 diabetes was performed using whole genome microarrays, the obtained data underwent quality control and were added to the pharmacogenomics study group. Neural networking programs are being tested using project data obtained during the project. A review article of pharmacogenomics of type 2 diabetes is being prepared.
Information published 29.10.2021.
Progress of the project:
1 November 2021 – 31 January 2022
The period from 01.11.2021. until 31.01.2022. were devoted to carry out the implementation of section 3 of the project: a sample of T2D has been prepared for the analysis of machine learning methods to obtain information on hidden subclusters within the T2D group. Genome-wide genotyping data from all genotyping batches were pooled and information about controls added (in several cases obtained using other genotyping chips) and information about non-overlapping SNPs was discarded, otherwise the machine learning algorithm would likely restore batches which correspond to genotyping with different chips at different times. The obtained pooled genotyping information was used for the analysis of T2D subsets, the results of which are still being collected. The information on the resulting subsets will be used to develop a prognostic model of disease progression for different groups of T2D patients. A preparation of review article of T2D pharmacogenetics is ongoing.
Information published 31.01.2022.
Progress of the project:
1 February 2022 – 30 April 2022
The period from 01.02.2022 until 30.04.2022 were used to carry out tasks for succesful closure of the project. Completion of required mobility period was done as e-mobility at Institute of Genomics, University of Tartu from 1 March till 20 April. During this e-mobility T2D clustering analysis was rounded up and deliverable 3.1 “Cluster analysis” prepared. Also during the e-mobility there was advanced training for machine learning implementation in data analysis and T2D model creation for Deliverable 3.3 provided.
Within the last three months of project article about T2D pharmacogenetics was completed. Deliverable 2.4 was prepared: overview of all GWAS results that were obtained during the project. Compilation of all project-endig related documents was finished.
Information published 29.04.2021.
Progress of the project
1 May 2022 – 31 July 2022
Studies of machine learning and clustering algorithms were conducted in order to perform quality patient data clustering and machine learning model creation. Literature studies for machine learning related publication were commenced.
Information published 29.07.2022.
Progress of the project
1 August 2022 – 31 October 2022
Machine learning algorithms were applied to the type 2 diabetes patients data, normalization and stratification methods were applied and machine learning models were created in order to classify patients and determine prognosis for common type two diabetes clinical outcomes: poorly controlled HbA1c level, type two diabetes vascular complications and mortality. Models were evaluated and their performance in our dataset described. Publication about machine learning, patient clustering and diabetes clinical outcomes prepared.
Information published 31.10.2022.