Molecular signature of soil organic matter under different land uses in the Lake Chaohu Basin.

The concentration and molecular composition of soil organic matter (SOM) are important factors in mitigation against climate change as well as providing other ecosystem services. Our quantitative understanding of how land use influences SOM molecular composition and associated turnover dynamics is limited, which underscores the need for high-throughput analytical approaches and molecular marker signatures to clarify this etiology. Combining a high-throughput untargeted mass spectrometry screening and molecular markers, we show that forest, farmland and urban land uses result in distinct molecular signatures of SOM in the Lake Chaohu Basin. Molecular markers indicate that forest SOM has abundant carbon contents from vegetation and condensed organic carbon, leading to high soil organic carbon (SOC) concentration. Farmland SOM has moderate carbon contents from vegetation, and limited content of condensed organic carbon, with SOC significantly lower than that of forest soils. Urban SOM has high abundance of condensed organic carbon markers due to anthropogenic activities but relatively low in markers from vegetation. Consistently, urban soils have the highest black carbon/SOC ratio among these land uses. Overall, our results suggested that the molecular signature of SOM varies significantly with land use in the Lake Chaohu Basin, influencing carbon dynamics. Our strategy of molecular fingerprinting and marker discovery is expected to enlighten further research on SOM molecular signatures and cycling dynamics.

Predicting apparent singlet oxygen quantum yields of dissolved black carbon and humic substances using spectroscopic indices.

Dissolved black carbon (DBC) is ubiquitous in aquatic systems, being an important subgroup of the dissolved organic matter (DOM) pool. Nevertheless, its aquatic photoactivity remains largely unknown. In this study, a range of spectroscopic indices of DBC and humic substance (HS) samples were determined using UV-Vis spectroscopy, fluorescence spectroscopy, and proton nuclear magnetic resonance. DBC can be readily differentiated from HS using spectroscopic indices. It has lower average molecular weight, but higher aromaticity and lignin content. The apparent singlet oxygen quantum yield (Φsinglet oxygen) of DBC under simulated sunlight varies from 3.46% to 6.13%, significantly higher than HS, 1.26%-3.57%, suggesting that DBC is the more photoactive component in the DOM pool. Despite drastically different formation processes and structural properties, the Φsinglet oxygen of DBC and HS can be well predicted by the same simple linear regression models using optical indices including spectral slope coefficient (S275-295) and absorbance ratio (E2/E3) which are proxies for the abundance of singlet oxygen sensitizers and for the significance of intramolecular charge transfer interactions. The regression models can be potentially used to assess the photoactivity of DOM at large scales with in situ water spectrophotometry or satellite remote sensing.

Binary Markov Random Fields and interpretable mass spectra discrimination.

For mass spectra acquired from cancer patients by MALDI or SELDI techniques, automated discrimination between cancer types or stages has often been implemented by machine learning algorithms. Nevertheless, these techniques typically lack interpretability in terms of biomarkers. In this paper, we propose a new mass spectra discrimination algorithm by parameterized Markov Random Fields to automatically generate interpretable classifiers with small groups of scored biomarkers. A dataset of 238 MALDI colorectal mass spectra and two datasets of 216 and 253 SELDI ovarian mass spectra respectively were used to test our approach. The results show that our approach reaches accuracies of 81% to 100% to discriminate between patients from different colorectal and ovarian cancer stages, and performs as well or better than previous studies on similar datasets. Moreover, our approach enables efficient planar-displays to visualize mass spectra discrimination and has good asymptotic performance for large datasets. Thus, our classifiers should facilitate the choice and planning of further experiments for biological interpretation of cancer discriminating signatures. In our experiments, the number of mass spectra for each colorectal cancer stage is roughly half of that for each ovarian cancer stage, so that we reach lower discrimination accuracy for colorectal cancer than for ovarian cancer.

Biomarker Signature Discovery from Mass Spectrometry Data.

Mass spectrometry based high throughput proteomics are used for protein analysis and clinical diagnosis. Many machine learning methods have been used to construct classifiers based on mass spectrometry data, for discrimination between cancer stages. However, the classifiers generated by machine learning such as SVM techniques typically lack biological interpretability. We present an innovative technique for automated discovery of signatures optimized to characterize various cancer stages. We validate our signature discovery algorithm on one new colorectal cancer MALDI-TOF data set, and two well-known ovarian cancer SELDI-TOF data sets. In all of these cases, our signature based classifiers performed either better or at least as well as four benchmark machine learning algorithms including SVM and KNN. Moreover, our optimized signatures automatically select smaller sets of key biomarkers than the black-boxes generated by machine learning, and are much easier to interpret.