Machine for shotgun metagenomic analysis | Information dissemination media for research and development TEGAKARI

A customer involved in research on the blood circulation system asked us for a machine for performing bioinformatics, mainly shotgun metagenomic analysis.The assumed conditions are:

・CPU: Focus on the number of cores
・ Memory: There are plans to expand in the future
・OS: Ubuntu 20.04
・Others: Bioconda3 pre-installed
・There is a plan to perform MRI image analysis using U-Net
・Budget: About 150 million yen

Based on the contents of the communication, we proposed the following configuration.

【Main Specifications】

CPU	Xeon Gold 6326 (2.90GHz 16 cores) x2
memory	128GB REG ECC
Storage 1	1TB SSD S-ATA
Storage 2	8TB HDD S-ATA
video	NVIDIA Geforce RTX4070 Ti
network	On Board (1000Base-T X2)
Housing + power supply	Tower type housing + 1200W
OS	Ubuntu 22.04
Others	Install CUDA Toolkit 11 Install Bioconda3

We are considering a configuration with as many cores as possible within the budget.
Considering the 64-core specification, the price will rise significantly and it will not fit in the budget, so we have reduced the total to 56 cores.

The memory is configured with future expansion in mind, but the original specification is to maximize the memory bandwidth by installing modules in all memory slots, so please note that the memory bandwidth is halved in the proposed state. please give me.

GPU has added one GPU that can be used for AI learning etc.
However, since it is not a GPU dedicated to calculation, please consider it as a necessary and sufficient product for an introductory class rather than for full-scale learning.

Regarding training using GPUs, there are issues such as the price of a single GPU being more than 100 million yen for higher-end products, and that multiple GPUs cannot be used in a general 100V environment.Considering such points, we are selecting GPUs that can be used for AI from the balance of usage and budget this time.
In addition, the memory capacity of the video card is also important when studying image systems, so if your budget allows, I think it would be better to choose an RTX A6000 48GB.

The configuration of this case study is based on the conditions given by the customer.
Please feel free to contact us even if you are considering different conditions from what is posted.

■FAQ

・What is shotgun metagenomic analysis?
Shotgun metagenomic analysis is a method of extracting DNA from environmental samples, randomly splitting it into fragments, and sequencing them to obtain genomic information.It is often used to investigate the diversity of microorganisms, and it is possible to obtain genome information of unknown microorganisms.Since there is no need to culture microorganisms, it is very fast and suitable for analysis of highly diverse environmental samples.

・What is U-Net?
U-Net is a deep learning architecture that demonstrates high performance in 2D image segmentation. U-Net consists of an encoder part and a decoder part, extracting image features using convolutional layers, lowering the resolution in the encoder part, and restoring the original resolution in the decoder part for image segmentation.It can accurately determine the position and shape of objects in an image, and is widely used in fields such as automatic analysis of medical images.