Advantages of Synthetic Noise and Machine Learning for Analyzing Radioecological Data Sets

Tritium: Its relevance, sources and impacts on non-human biota

Tritium (³H) is a radioactive isotope of hydrogen that is abundantly released from nuclear industries. It is extremely mobile in the environment and in all biological systems, representing an increasing concern for the health of both humans and non-human biota (NHB). The present review examines the sources and characteristics of tritium in the environment, and evaluates available information pertaining to its biological effects at different levels of biological organisation in NHB. Despite an increasing number of publications in the tritium radiobiology field, there exists a significant disparity between data available for the different taxonomic groups and species, and observations are heavily biased towards marine bivalves, fish and mammals (rodents). Further limitations relate to the scarcity of information in the field relative to the laboratory, and lack of studies that employ forms of tritium other than tritiated water (HTO). Within these constraints, different responses to HTO exposure, from molecular to behavioural, have been reported during early life stages, but the potential transgenerational effects are unclear. The application of rapidly developing "omics" techniques could help to fill these knowledge gaps and further elucidate the relationships between molecular and organismal level responses through the development of radiation specific adverse outcome pathways (AOPs). The use of a greater diversity of keystone species and exposures to multiple stressors, elucidating other novel effects (e.g., by-stander, germ-line, transgenerational and epigenetic effects) offers opportunities to improve environmental risk assessments for the radionuclide. These could be combined with artificial intelligence (AI) including machine learning (ML) and ecosystem-based approaches.

Machine learning analysis of 137Cs contamination of terrestrial plants after the Fukushima accident using the random forest algorithm

Radioactive contamination of terrestrial plants was extensively investigated and quantitatively modeled after the Fukushima nuclear power plant accident. This phenomenon, which is important for ecosystem functioning and protection of human health, is influenced by multiple factors, including plant species, time after the accident, and climate. Machine learning algorithms such as random forests (RF) have a record of strong performance on large multi-dimensional data sets, but, to our knowledge, combined data on post-Fukushima plant contamination with radionuclides were not yet subjected to a machine learning analysis. Here we performed such analysis on two large published data sets: (1) ¹³⁷Cs activity concentrations in four common Japanese forest tree species. (2) Plant/soil ¹³⁷Cs concentration ratios in multiple perennial plant species. The goal was to show the usefulness of machine learning for identifying and quantifying the main trends of ¹³⁷Cs contamination in terrestrial plants. Each data set was split randomly into training and testing parts, RF was fitted and tuned on the training parts, and its performance was assessed on the testing parts by three metrics: coefficient of determination (R²), root mean squared error, and mean absolute error. Synthetic noise variables and the Boruta algorithm were used in a customized procedure to identify the most important predictor variables, which consistently outperformed random noise. Good agreement between observations and RF predictions (e.g. R²∼0.9 on testing data) was obtained on both data sets. The effects of the most important predictors (e.g. time after the accident, ¹³⁷Cs land contamination level, and plant species) and interactions between them were quantified by partial dependence plots. These results of machine learning analyses of large data collections can help to complement previous modeling efforts, and to clarify the patterns of ¹³⁷Cs contamination of plants after the Fukushima accident.

New Approaches for Quantitative Reconstruction of Radiation Dose in Human Blood Cells

In the event of a nuclear attack or large-scale radiation event, there would be an urgent need for assessing the dose to which hundreds or thousands of individuals were exposed. Biodosimetry approaches are being developed to address this need, including transcriptomics. Studies have identified many genes with potential for biodosimetry, but, to date most have focused on classification of samples by exposure levels, rather than dose reconstruction. We report here a proof-of-principle study applying new methods to select radiation-responsive genes to generate quantitative, rather than categorical, radiation dose reconstructions based on a blood sample. We used a new normalization method to reduce effects of variability of signal intensity in unirradiated samples across studies; developed a quantitative dose-reconstruction method that is generally under-utilized compared to categorical methods; and combined these to determine a gene set as a reconstructor. Our dose-reconstruction biomarker was trained using two data sets and tested on two independent ones. It was able to reconstruct dose up to 4.5 Gy with root mean squared error (RMSE) of ± 0.35 Gy on a test dataset using the same platform, and up to 6.0 Gy with RMSE of ± 1.74 Gy on a test set using a different platform.

Chronic gamma radiation resistance in fungi correlates with resistance to chromium and elevated temperatures, but not with resistance to acute irradiation

Exposure to chronic ionizing radiation (CIR) from nuclear power plant accidents, acts of terrorism, and space exploration poses serious threats to humans. Fungi are a group of highly radiation-resistant eukaryotes, and an understanding of fungal CIR resistance mechanisms holds the prospect of protecting humans. We compared the abilities of 95 wild-type yeast and dimorphic fungal isolates, representing diverse Ascomycota and Basidiomycota, to resist exposure to five environmentally-relevant stressors: CIR (long-duration growth under 36 Gy/h) and acute (10 kGy/h) ionizing radiation (IR), heavy metals (chromium, mercury), elevated temperature (up to 50 °C), and low pH (2.3). To quantify associations between resistances to CIR and these other stressors, we used correlation analysis, logistic regression with multi-model inference, and customized machine learning. The results suggest that resistance to acute IR in fungi is not strongly correlated with the ability of a given fungal isolate to grow under CIR. Instead, the strongest predictors of CIR resistance in fungi were resistance to chromium (III) and to elevated temperature. These results suggest fundamental differences between the mechanisms of resistance to chronic and acute radiation. Convergent evolution towards radioresistance among genetically distinct groups of organisms is considered here.

Transcriptomic analysis reveals the relationship of melanization to growth and resistance to gamma radiation in Cryptococcus neoformans

The pathogenic fungus Cryptococcus neoformans produces melanin within its cell wall for infection and resistance against external stresses such as exposure to UV, temperature fluctuations and reactive oxygen species. It has been reported that melanin may also protect cells from ionizing radiation damage, against which C. neoformans is extremely resistant. This has tagged melanin as a potential radioprotective biomaterial. Here, we report the effect of melanin on the transcriptomic response of C. neoformans to gamma radiation. We did not observe a substantial protective effect of melanin against gamma radiation, and the general gene expression patterns in irradiated cells were independent of the presence of melanin. However, melanization itself dramatically altered the C. neoformans transcriptome, primarily by repressing genes involved in respiration and cell growth. We suggest that, in addition to providing a physical and chemical barrier against external stresses, melanin production alters the transcriptional landscape of C. neoformans with the result of increased resistance to uncertain environmental conditions. This observation demonstrates the importance of the melanization process in understanding the stress response of C. neoformans and for understanding fungal physiology.

Modeling species richness and abundance of phytoplankton and zooplankton in radioactively contaminated water bodies

Water bodies polluted by the Mayak nuclear plant in Russia provide valuable information on multi-generation effects of radioactive contamination on freshwater organisms. For example, lake Karachay was probably the most radioactive lake in the world: its water contained ∼2 × 10⁷ Bq/L of radionuclides and estimated dose rates to plankton exceeded 5 Gy/h. We performed quantitative modeling of radiation effects on phytoplankton and zooplankton species richness and abundance in Mayak-contaminated water bodies. Due to collinearity between radioactive contamination, water body size and salinity, we combined these variables into one (called HabitatFactors). We employed a customized machine learning approach, where synthetic noise variables acted as benchmarks of predictor performance. HabitatFactors was the only predictor that outperformed noise variables and, therefore, we used it for parametric modeling of plankton responses. Best-fit model predictions suggested 50% species richness reduction at HabitatFactors values corresponding to dose rates of 10⁴-10⁵ μGy/h for phytoplankton, and 10³-10⁴ μGy/h for zooplankton. Under conditions similar to those in lake Karachay, best-fit models predicted 81-98% species richness reductions for various taxa (Cyanobacteria, Bacillariophyta, Chlorophyta, Rotifera, Cladocera and Copepoda), ∼20-300-fold abundance reduction for total zooplankton, but no abundance reduction for phytoplankton. Rotifera was the only taxon whose fractional abundance increased with contamination level, reaching 100% in lake Karachay, but Rotifera species richness declined with contamination level, as in other taxa. Under severe radioactive and chemical contamination, one species of Cyanobacteria (Geitlerinema amphibium) dominated phytoplankton, and rotifers from the genus Brachionus dominated zooplankton. The modeling approaches proposed here are applicable to other radioecological data sets. The results provide quantitative information and easily interpretable model parameter estimates for the shapes and magnitudes of freshwater plankton responses to a wide range of radioactive contamination levels.