This study aims to provide insights into the COVID-19-related communication on Twitter in the Republic of Croatia. For that purpose, we developed an NL-based framework that enables automatic analysis of a large dataset of tweets in the Croatian language. We collected and analysed 206,196 tweets related to COVID-19 and constructed a dataset of 10,000 tweets which we manually annotated with a sentiment label. We trained the Cro-CoV-cseBERT language model for the representation and clustering of tweets. Additionally, we compared the performance of four machine learning algorithms on the task of sentiment classification. After identifying the best performing setup of NLP methods, we applied the proposed framework in the task of characterisation of COVID-19 tweets in Croatia. More precisely, we performed sentiment analysis and tracked the sentiment over time. Furthermore, we detected how tweets are grouped into clusters with similar themes across three pandemic waves. Additionally, we characterised the tweets by analysing the distribution of sentiment polarity (in each thematic cluster and over time) and the number of retweets (in each thematic cluster and sentiment class). These results could be useful for additional research and interpretation in the domains of sociology, psychology or other sciences, as well as for the authorities, who could use them to address crisis communication problems.

Retweet prediction is an important task in the context of various problems, such as information spreading analysis, automatic fake news detection, social media monitoring, etc. In this study, we explore retweet prediction based on heterogeneous data sources. In order to classify a tweet according to the number of retweets, we combine features extracted from the multilayer network and text. More specifically, we introduce a multilayer framework for the multilayer network representation of Twitter. This formalism captures different users’ actions and complex relationships, as well as other key properties of communication on Twitter. Next, we select a set of local network measures from each layer and construct a set of multilayer network features. We also adopt a BERT-based language model, namely Cro-CoV-cseBERT, to capture the high-level semantics and structure of tweets as a set of text features. We then trained six machine learning (ML) algorithms: random forest, multilayer perceptron, light gradient boosting machine, category-embedding model, neural oblivious decision ensembles, and an attentive interpretable tabular learning model for the retweet-prediction task. We compared the performance of all six algorithms in three different setups: with text features only, with multilayer network features only, and with both feature sets. We evaluated all the setups in terms of standard evaluation measures. For this task, we first prepared an empirical dataset of 199,431 tweets in Croatian posted between 1 January 2020 and 31 May 2021. Our results indicate that the prediction model performs better by integrating multilayer network features with text features than by using only one set of features.

Disrupted in Schizophrenia 1 (DISC1) is a scaffold protein implicated in major mental illnesses including schizophrenia, with a significant negative impact on social life. To investigate if DISC1 affects social interactions in Drosophila melanogaster, we created transgenic flies with second or third chromosome insertions of the human full-length DISC1 (hflDISC1) gene fused to a UAS promotor (UAS-hflDISC1). Initial characterization of the insertion lines showed unexpected endogenous expression of the DISC1 protein that led to various behavioral and neurochemical phenotypes. Social interaction network (SIN) analysis showed altered social dynamics and organizational structures. This was in agreement with the altered levels of the locomotor activity of individual flies monitored for 24 h. Together with a decreased ability to climb vertical surfaces, the observed phenotypes indicate altered motor functions that could be due to a change in the function of the motor neurons and/or central brain. The changes in social behavior and motor function suggest that the inserted hflDISC1 gene influences nervous system functioning that parallels symptoms of DISC1-related mental diseases in humans. Furthermore, neurochemical analyses of transgenic lines revealed increased levels of hydrogen peroxide and decreased levels of glutathione, indicating an impact of DISC1 on the dynamics of redox regulation, similar to that reported in transgenic mammals. Future studies are needed to address the localization of DISC1 expression and to address how the redox parameter changes correlate with the observed behavioral changes.

Addiction is a multifactorial biological and behavioral disorder that is studied using animal models, based on simple behavioral responses in isolated individuals. A couple of decades ago it was shown thatDrosophila melanogastercan serve as a model organism for behaviors related to alcohol, nicotine and cocaine (COC) addiction. Scoring of COC-induced behaviors in a large group of flies has been technologically challenging, so we have applied a local, middle and global level of network-based analyses to study social interaction networks (SINs) among a group of 30 untreated males compared to those that have been orally administered with 0.50 mg/mL of COC for 24 hours. In this study, we have confirmed the previously described increase in locomotion upon COC feeding. We have isolated new network-based measures associated with COC, and influenced by group on the individual behavior. COC fed flies showed a longer duration of interactions on the local level, and formed larger, more densely populated and compact, communities at the middle level. Untreated flies have a higher number of interactions with other flies in a group at the local level, and at the middle level, these interactions led to the formation of separated communities. Although the network density at the global level is higher in COC fed flies, at the middle level the modularity is higher in untreated flies. One COC specific behavior that we have isolated was an increase in the proportion of individuals that do not interact with the rest of the group, considered as the individual difference in COC induced behavior and/or consequence of group influence on individual behavior. Our approach can be expanded on different classes of drugs with the same acute response as COC to determine drug specific network-based measures and could serve as a tool to determinate genetic and environmental factors that influence both drug addiction and social interaction.

Fragile X syndrome (FXS), the most common monogenic cause of inherited intellectual disability and autism spectrum disorder, is caused by a full mutation (>200 CGG repeats) in the Fragile X Messenger Ribonucleoprotein 1 (FMR1) gene. Individuals with FXS experience various challenges related to social interaction (SI). Animal models, such as the Drosophila melanogaster model for FXS where the only ortholog of human FMR1 (dFMR1) is mutated, have played a crucial role in the understanding of FXS. The aim of this study was to investigate SI in the dFMR1^B55 mutants (the groups of flies of both sexes simultaneously) using the novel Drosophila Shallow Chamber and a Python data processing pipeline based on social network analysis (SNA). In comparison with wild-type flies (w¹¹¹⁸), SNA analysis in dFMR1^B55 mutants revealed hypoactivity, fewer connections in their networks, longer interaction duration, a lower ability to transmit information efficiently, fewer alternative pathways for information transmission, a higher variability in the number of interactions they achieved, and flies tended to stay near the boundaries of the testing chamber. These observed alterations indicate the presence of characteristic strain-dependent social networks in dFMR1^B55 flies, commonly referred to as the group phenotype. Finally, combining novel research tools is a valuable method for SI research in fruit flies.