The influence of the inclusion of biological knowledge in statistical methods to integrate multi-omics data

Understanding the relationships among biomolecules and how these relationships change between healthy and disease states is an important question in modern biology and medicine. The advances in high-throughput techniques has led to the explosion of biological data available for analysis, allowing researchers to investigate multiple molecular layers (i.e. omics data) together. The classical statistical methods could not address the challenges of combining multiple data types, leading to the development of ad hoc methodologies, which however depend on several factors. Among those, it is important to consider whether “prior knowledge†on the inter-omics relationships is available for integration. To address this issue, we thus focused on different approaches to perform three-omics integration: supervised (prior knowledge is available), unsupervised and semi-supervised. With the supervised integration of DNA methylation, gene expression and protein levels from adipocytes we observed coordinated significant changes across the three omics in the last phase of adipogenesis. However, in most cases, interactions between different molecular layers are complex and unknown: we explored unsupervised integration methods, showing that their results are influenced by method choice, pre-processing, number of integrated data types and experimental design. The strength of the inter-omics signal and the presence of noise are also proven as relevant factors. Since the inclusion of prior knowledge can highlight the former while decreasing the influence of the latter, we proposed a semi-supervised approach, showing that the inclusion of knowledge about inter-omics interactions increases the accuracy of unsupervised methods when solving the problem of sample classification. Word retrieval deficits are a common problem in patients with stroke-induced brain damage. While the complete recovery of language in chronic aphasia is rare, patients’ naming ability can be significantly improved by speech therapy. A growing number of neuroimaging studies have tried to pinpoint the neural mechanisms associated with successful outcome of naming treatment. Although the picture is complicated by the large variability in lesion and patient characteristics, some advances have been made in this direction. However, one piece of the puzzle is missing. Namely, the brain mechanisms supporting naming practice in the healthy brain have not received enough attention. Yet, understanding these mechanisms is crucial for teasing them apart from functional reorganization resulting from brain damage. In this thesis I present an fMRI study of intensive naming practice in healthy monolingual adults. Subjects were trained on naming objects and actions from pictures for ten consecutive days and were scanned twice, before and after training. Training of noun and verb production was associated with a similar set of BOLD activation changes, encompassing both anterior and posterior regions of the left hemisphere. Interestingly, while the anterior regions (posterior inferior frontal gyrus, anterior insula, anterior cingulate cortex) showed significant practice-related BOLD decreases, activity in the posterior regions (precuneus, posterior cingulate cortex, angular gyrus, posterior middle temporal gyrus) increased following training. We argue that while the activation decreases were likely associated with the facilitation of different aspects of word production, the greater engagement of parietal and temporal cortices after training potentially reflects retrieval of knowledge pertaining to trained items from episodic (and potentially semantic) memory. Additionally, using the data from the pre-training fMRI session, we investigated the putative dissociation between neural representations of nouns and verbs. In line with previous reports, we found that action naming recruited bilateral lateral occipitotemporal and posterior parietal and inferior frontal cortices (predominantly left-lateralized) to a greater degree than object naming. Our second experiment aimed to further scrutinize categorical distinctions in lexical representation by tackling the effects of verb transitivity. Corroborating previous findings, we observed greater recruitment of several regions in a fronto-temporo-parietal network for transitive verbs. However, in alternative to previous reports that accounted for the results in terms of linguistic processes, we interpreted the observed effects in terms of perceptual and conceptual differences between transitive and intransitive verbs. We outline the major questions that should be addressed by future studies.