Edelweiss Chemical Science Journal (ISSN 2641-7383)

Editorial :

Experience, Data-Driven and Artificial Intelligence in Social (Fire) and Chemical Technology

Full-Text

Introduction

Realization of society 5.0 in fire and disaster prevention activities is one of intensive goals of Japanese government developing fire technology [1]. Improvement of new equipment and materials for disaster response utilizing AI and ICT should be developed according to social requirements. Efforts to predict earthquake, pour, flood, etc. through AI analysis of data collected from past disasters must continue. In parallel with such elaboration of disaster prediction, it is necessary to proceed with preparations for prompt and accurate provision of disaster information during emergency situations and support for rebuilding lives post disaster.

Discussion 

Recently, Spring-8 (large synchrotron facility in Japan) has developed a program Deep Centering that automatically detects protein crystal samples in X-ray crystal structure analysis by image analysis using deep learning [2]. The result of this research realizes automatic positioning of protein crystals and is expected to be applied to fully automatic data collection and automatic structural analysis. In X-ray crystal structure analysis, a crystal centering is carried out position the sample crystal in the X-ray optical path irrespective of direction of rotation. Conventionally, this work is carried out mainly by the user of the beam line or by detecting the position of the crystal by irradiating the crystal with a weakened X-ray. This has made it possible to save labor and avoid radiation damage caused by X-ray. 

According to bad examples reported by Akitsu [3], the normal procedure for manually centering a single crystal is described as follows. Firstly, a single crystal is attached to move sideways so that the center of the single crystal overlaps the center line (usually toward the center) in the direction seen with the microscope. Now axis is rotated (Φ) 180 degrees. So, in the direction being looked at the microscope, it is moved sideways so that the center of the single crystal overlaps the center line (usually towards the center), and repeat several times (but many times due to vertical misalignment). 

Then, axis is rotated (Φ) 90 degrees. Therefore, in the direction being seen with the microscope, the single crystal is moved sideways so that the center of the single crystal overlaps the center line (usually toward the center). And axis is rotated (Φ) 180 degrees. Therefore, in a direction as viewed with a microscope, the lateral movement is repeated several times so that the center of the single crystal is overlapped the center line (usually toward the center). Finally, axis is rotated every (Φ) 90 degrees to be confirmed that the center of the single crystal is overlapped the center line. In this way, the final correct position is assumed to be on the inner side than the initial state (the normal deviation is maximized), and the deviation from the center line of the single crystal is adjusted to be smaller. 

If the center of the single crystal is shifted from the aiming center line from the beginning. Which is difficult to be aligned the center position by this method only by rotating the Φ axis. Although I have read the APEX manual by the Department of Chemistry at Purdue University, where Dr. Negishi was awarded the Nobel Prize in Chemistry, there was a figure very similar to that Asahi Beers object on P11 [4-6].Figure 1, however, exhibits terrible example of failure alignment by a novice in my group.

Figure 1: Failure of single crystal experiments; [left] Bad alignment, [middle] Poor resolution, and [right] unusual data.

Conclusion 

Even though researches based on data are good at interpolation, it is said that prediction beyond experience is not good. In particular, in Japan, there are many excuses for out of scope (Sotei-gai in Japanese) when accident or disaster occur actually. Therefore, I think what I have done after seeing immatures Sotei-gai poor experiments. 

References

1.     Details of the technologies covered by the important R and D program and the time to achieve the target results.

2.     Press release Development of an Automatic Crystal Centering Program Using Deep Learning-A Fully Automatic Crystal Structure Analysis of Proteins.

3.     Akitsu T. Crystallography (2019) InTech open, Croatia, pp: 1-4.

4.     Standard operating procedure-bruker quest diffractometer with sealed tube molybdenum source.

5.     Ei-ichi Negishi. Nobel Laureate (2010).

6.     What do the gold objects next to the Asahi Breweries head office building in Asakusa represent? 

^*Corresponding author

Takashiro Akitsu, Professor, Department of Chemistry, Faculty of Science, Tokyo University of Science, 1-3 Kagurazaka, Shinjuku-ku, Tokyo 162-8601, Japan, E-mail: akitsu2@rs.tus.ac.jp 

Citation

Akitsu T and Onami Y. Experience, data-driven and artificial intelligence in social (fire) and chemical technology (2019) Edelweiss Chem Sci J 2: 45-46.

Keywords

Artificial intelligence, Information and communications technology, Crystallography, Out-of-scope prediction.