Project Details
Projekt
Proteomic fingerprinting for species identification – discriminatory power and optimal analyses procedures for integrated molecular and morphological datasets in zooplankton biodiversity assessments
Applicant
Dr. Jasmin Renz-Gehnke
Subject Area
Ecology and Biodiversity of Animals and Ecosystems, Organismic Interactions
Term
since 2017
Project identifier
Deutsche Forschungsgemeinschaft (DFG) - Project number 349621075
Zooplankton communities are ‘rapid responders’ to changing biological and physical conditions, and are as such useful indicators of environmental variation and climate change. Many of them are characterised by a fast turnover of species composition, high diversity and numerous specimens that are challenging to identify based on morphological characteristics. Thus, there is a crucial need for a fast, easy-to-use and inexpensive assessment tool providing comprehensive baselines and reliable monitoring of species diversity. The application of species-specific proteomic mass profiles measured by MALDI-TOF MS (matrix-assisted laser/desorption ionization time-of-flight mass spectrometry) has strong potential to serve as such a tool. The method is well-established in microbiology and several proof of concept studies revealed its applicability for invertebrates. However, most of these studies focused on a limited group of taxa, did not account for potential phenotypic plasticity or test for discriminatory stability using more complex and diverse datasets. Yet, these are issues that need to be addressed when using this method in multi-taxa community studies. Within this project we will test for the first time the applicability of proteomic fingerprinting for zooplankton species discrimination and optimise data processing for monitoring and biodiversity assessments respectively, using the North Sea as a case study. In detail, we will determine the inter- and intraspecific variability of proteomic profiles of approx. 100-150 zooplankton species (from 13 different phyla), test whether closely related species can be reliably discriminated and assess the impacts of community complexity on discrimination performance based on proteomic features. Using these findings we will explore, which (semi-)supervised machine learning procedure will be best suited for species classification in zooplankton with special emphasis on the implementation of novelty detection (e.g. for monitoring). We will specifically investigate the impacts of feature selection, compare performance of random forest models, support vector machine and neural networks and evaluate whether an integration of proteomic and non-proteomic features will increase classification success. Furthermore, we will establish an unsupervised machine learning procedure for species discrimination (e.g. in biodiversity assessments) using integrated molecular and morphological datasets. We will apply this method to the highly diverse zooplankton community from the Bermuda Atlantic time-series study in the Sargasso Sea. Finally, we will develop a strategy for FAIR (Findable, Accessible, Interoperable, Reusable) data storage of integrative data sets on proteomic profiles, taxonomy and genetic barcodes. This study will fundamentally contribute to our general understanding of method validity and also of pitfalls using MALDI-TOF MS for analysis of more complex metazoan species communities.
DFG Programme
Priority Programmes
