|
ИСТИНА |
Войти в систему Регистрация |
ИПМех РАН |
||
Unsupervised neural network model helps to attribute particular stages of chlorophyll a fluorescence transientтезисы доклада
- Авторы: Sergey Khruschev, Tatiana Plyusnina, Galina Riznichenko, Rubin Andrey B.
- Сборник: 10th International Conference “Photosynthesis and Hydrogen Energy Research for Sustainability-2019", June 23 – 28, 2019, Saint Petersburg, Russia, Abstracts, ISBN 978-5-4386-1743-3
- Тезисы
- Год издания: 2019
- Место издания: Komarov Botanical Institute of the Russian Academy of Sciences St. Petersburg
- Первая страница: 97
- Последняя страница: 97
- Аннотация: Analysis of chlorophyll a fluorescence transient upon illumination after dark adaptation is widely used to assess the activity of primary photosynthetic processes. Modern fluorometric equipment allows detailed recording of induction curves with time resolution of microseconds. For green algae and higher plants fast fluorescence rise kinetics is generally described as three-step (O-J-I-P), however comparison of transients recorded under nutrient surplus and deficiency reveal sequence of much more bands (O-L-K-J-I-H-G-P) [1]. It is commonly accepted that individual phases of the induction curve depict various stages of the electron transfer in the electron transport chain (or exciton transfer between Photosystem II reaction centers); however attribution of the phases to specific processes is still under discussion. Construction of a robust method for quantitative description of the transient in terms of characteristic times and amplitudes of phases is essential for this task. Notion of photosystem as a huge complex containing several redox centers acting in chain implies that individual phases should be considered as exponential functions, each of which represent some set of processes (mostly related to electron transfer). Characteristic times of these exponents may be rather close to each other, and they vary with temperature, light intensity, and physiological state of the organism, making unequivocal determination of numeric parameters from experimental data a sophisticated task. To deal with this ambiguity, we suggest a mathematical method for simultaneous analysis of a set of fluorescence transients recorded under varying conditions. The use of an unsupervised neural network model allows finding minimal set of exponents suitable to describe a given set of experimental data. It was shown that L-band (see [1]) related to energetic coupling of PS2 RCs can be identified as a distinct exponential component with characteristic time of about 100 μs or below, having negative contribution to the fluorescence rise under normal conditions (thus leading to ‘sigmoidity’ of the early stage of the transient), and positive contribution in stress conditions (i.e., nitrogen or sulfur deficiency, high illumination, etc.). This work was partially supported by RFBR project №17-04-00676.
- Добавил в систему: Хрущев Сергей Сергеевич