Gender Identification and Classification of Drosophila melanogaster Flies Using Machine Learning Techniques

Drosophila melanogaster is an important genetic model organism used extensively in medical and biological studies. About 61% of known human genes have a recognizable match with the genetic code of Drosophila flies, and 50% of fly protein sequences have mammalian analogues. Recently, several investigations have been conducted in Drosophila to study the functions of specific genes exist in the central nervous system, heart, liver, and kidney. The outcomes of the research in Drosophila are also used as a unique tool to study human-related diseases. This article presents a novel automated system to classify the gender of Drosophila flies obtained through microscopic images (ventral view). The proposed system takes an image as input and converts it into grayscale illustration to extract the texture features from the image. Then, machine learning (ML) classifiers such as support vector machines (SVM), Naive Bayes (NB), and K -nearest neighbour (KNN) are used to classify the Drosophila as male or female. The proposed model is evaluated using the real microscopic image dataset, and the results show that the accuracy of the KNN is 90%, which is higher than the accuracy of the SVM classifier.

The transcriptomic signature of physiological trade-offs caused by larval overcrowding in Drosophila melanogaster

Intraspecific competition at the larval stage is an important ecological factor affecting life-history, adaptation and evolutionary trajectory in holometabolous insects. However, the molecular pathways and physiological trade-offs underpinning these ecological processes are poorly characterised. We reared Drosophila melanogaster at three egg densities (5, 60 and 300 eggs/ml) and sequenced the transcriptomes of pooled third-instar larvae. We also examined emergence time, egg-to-adult viability, adult mass and adult sex-ratio at each density. Medium crowding had minor detrimental effects on adult phenotypes compared to low density and yielded 24 differentially expressed genes (DEGs) including several chitinase enzymes. In contrast, high crowding had substantial detrimental effects on adult phenotypes and yielded 2107 DEGs. Among these, upregulated gene sets were enriched in sugar, steroid and amino acid metabolism as well as DNA replication pathways, whereas downregulated gene sets were enriched in ABC transporters, Taurine, Toll/Imd signalling and P450 xenobiotics metabolism pathways. Overall, our findings show that larval overcrowding has a large consistent effect on several molecular pathways (i.e., core responses) with few pathways displaying density-specific regulation (i.e., idiosyncratic responses). This provides important insights into how holometabolous insects respond to intraspecific competition during development.

Bugs on Drugs: A Drosophila melanogaster Gut Model to Study In Vivo Antibiotic Tolerance of E. coli

With an antibiotic crisis upon us, we need to boost antibiotic development and improve antibiotics’ efficacy. Crucial is knowing how to efficiently kill bacteria, especially in more complex in vivo conditions. Indeed, many bacteria harbor antibiotic-tolerant persisters, variants that survive exposure to our most potent antibiotics and catalyze resistance development. However, persistence is often only studied in vitro as we lack flexible in vivo models. Here, I explored the potential of using Drosophila melanogaster as a model for antimicrobial research, combining methods in Drosophila with microbiology techniques: assessing fly development and feeding, generating germ-free or bacteria-associated Drosophila and in situ microscopy. Adult flies tolerate antibiotics at high doses, although germ-free larvae show impaired development. Orally presented E. coli associates with Drosophila and mostly resides in the crop. E. coli shows an overall high antibiotic tolerance in vivo potentially resulting from heterogeneity in growth rates. The hipA7 high-persistence mutant displays an increased antibiotic survival while the expected low persistence of ΔrelAΔspoT and ΔrpoS mutants cannot be confirmed in vivo. In conclusion, a Drosophila model for in vivo antibiotic tolerance research shows high potential and offers a flexible system to test findings from in vitro assays in a broader, more complex condition.