Search bioRxiv⌕ Search

Biology subjects

Schwarz, S.

Publications and source records attributed to Schwarz, S..

21 records · Page 2Linked to original sources

Avoiding pitfalls: Trace conditioning and rapid aversive learning during route navigation in desert ants

The ability of bees and ants to learn long visually guided routes in complex environments is perhaps one of the most spectacular pieces of evidence for the impressive power of their small brains. While flying bees can visit flowers in an optimised sequence over kilometres, walking ants can precisely recapitulate routes of up to a hundred metres in complex environments. It is clear that route following depends largely on learnt visual information and we have good idea how views can guide individuals along them, however little is known about the mechanisms that control route learning and development. Here we show that ants in natural environments can actively learn a route detour to avoid a pit trap and that this depends on a process analogous to aversive trace conditioning. Views experienced before falling into the trap become associated with the ensuing negative outcome and thus trigger salutary turns on the subsequent trip. This drives the ants to orient away from the goal direction and avoid the trap. If the pit is avoided, the novel views experienced during the detour become positively reinforced and the new route crystallises. We discuss how such an interplay between appetitive and aversive memories might be implemented in insect neural circuitry.

animal behavior and cognition↗

Human-level recognition of blast cells in acute myeloid leukemia with convolutional neural networks

Reliable recognition of malignant white blood cells is a key step in the diagnosis of hematologic malignancies such as Acute Myeloid Leukemia. Microscopic morphological examination of blood cells is usually performed by trained human examiners, making the process tedious, time-consuming and hard to standardise. We compile an annotated image dataset of over 18,000 white blood cells, use it to train a convolutional neural network for leukocyte classification, and evaluate the networks performance. The network classifies the most important cell types with high accuracy. It also allows us to decide two clinically relevant questions with human-level performance, namely (i) if a given cell has blast character, and (ii) if it belongs to the cell types normally present in non-pathological blood smears. Our approach holds the potential to be used as a classification aid for examining much larger numbers of cells in a smear than can usually be done by a human expert. This will allow clinicians to recognize malignant cell populations with lower prevalence at an earlier stage of the disease.

pathology↗

Multi-omics approach identifies novel pathogen-derived prognostic biomarkers in patients with Pseudomonas aeruginosa bloodstream infection

Pseudomonas aeruginosa is a human pathogen that causes health-care associated blood stream infections (BSI). Although P. aeruginosa BSI are associated with high mortality rates, the clinical relevance of pathogen-derived prognostic biomarker to identify patients at risk for unfavorable outcome remains largely unexplored. We found novel pathogen-derived prognostic biomarker candidates by applying a multi-omics approach on a multicenter sepsis patient cohort. Multi-level Cox regression was used to investigate the relation between patient characteristics and pathogen features (2298 accessory genes, 1078 core protein levels, 107 parsimony-informative variations in reported virulence factors) with 30-day mortality. Our analysis revealed that presence of the helP gene encoding a putative DEAD-box helicase was independently associated with a fatal outcome (hazard ratio 2.01, p = 0.05). helP is located within a region related to the pathogenicity island PAPI-1 in close proximity to a pil gene cluster, which has been associated with horizontal gene transfer. Besides helP, elevated protein levels of the bacterial flagellum protein FliL (hazard ratio 3.44, p < 0.001) and of a bacterioferritin-like protein (hazard ratio 1.74, p = 0.003) increased the risk of death, while high protein levels of a putative aminotransferase were associated with an improved outcome (hazard ratio 0.12, p < 0.001). The prognostic potential of biomarker candidates and clinical factors was confirmed with different machine learning approaches using training and hold-out datasets. The helP genotype appeared the most attractive biomarker for clinical risk stratification due to its relevant predictive power and ease of detection.

microbiology↗