Bertis Publishes Findings on ‘DeepMRM’, a Targeted Proteomics Data Interpretation Software Package > BERTIS NEWS


NEWSROOM

Read latest news from Bertis

BERTIS NEWS

BERTIS NEWS

Bertis Publishes Findings on ‘DeepMRM’, a Targeted Proteomics Data Interpretation Software Package

Date 2023.07.12

Hit 1,043

News

Bertis Publishes Findings on ‘DeepMRM’, a Targeted Proteomics

Data Interpretation Software Package in International Journal

 

-Publish findings on assessing the performance and accuracy of the self-developed ‘DeepMRM’ in ‘Cell Reports Methods’, an international journal, on July 12

-According to the results of the comparative test on the performance of discovering protein candidates, DeepMRM has higher accuracy than existing software and has the lowest error percentage

-Promote high-speed analysis in the clinical environment and improve reproducibility and scalability of targeted proteomics analysis

 

On July 12, Bertis, a company specializing in the development of proteomics-based precision medicine technology (led by CEOs Dong-young Noh and Seung-man Han), announced that it had presented its findings on assessing the performance of ‘DeepMRM’, a software package based on deep learning algorithms for object detection, in ‘Cell Reports Methods’.

 

‘DeepMRM’ is a software package which the Bertis research team has developed by employing deep learning technology with the aim of more efficiently and accurately conducting MRM (Multiple Reaction Monitoring), a representative analysis method for discovering targeted protein candidates. With an existing community standard tool for MRM, researchers often devote substantial time to manual data interpretation, which hinders the transferability, reproducibility, and scalability of this approach. Deep MRM addresses these limitations by utilizing deep learning technology for object detection.

 

The Bertis Machine Learning team and its U.S. subsidiary Bertis Bioscience team utilized public MRM dataset to test the performance of DeepMRM against the most widely used Skyline. The results demonstrated that DeepMRM outperformed Skyline in both correlation coefficient between quantification values and actual values of analyzed proteomes, and MAAPE (Mean Arctangent Absolute Percentage Error), which measures average error values. Even when applying the mProphet algorithm, a filter that allows skyline to make the results more refined, DeepMRM's results were more accurate.>

 

32b59d5b02ceeb96e1ec64006239ce96_1689814108_4153.jpg
 

<Image: Table comparing the performances of DeepMRM and existing software>

 

With its superior performance, DeepMRM not only reduces proteomic analysis time in clinical research, but also improves the reproducibility of analysis results, increasing the reliability of results. In addition, thanks to its high scalability, DeepMRM can analyze other types of proteins, such as PRM (Parallel Reaction Monitoring) and DIA (Data-Independent Acquisition) data, in addition to MRM data.



 32b59d5b02ceeb96e1ec64006239ce96_1689814241_9921.jpg

<Image: Process for analyzing targeted proteomics using DeepMRM>

 

Jungkap Park, Director of Machine Learning team at Bertis, said, “DeepMRM is an innovative analysis tool that can dramatically increase data throughput in Bertis' major research areas - multi-biomarker diagnostic solutions and analysis services. Bertis will continue its R&D efforts towards AI-powered analytical tools to develop a unique platform which enables the diagnosis and treatment of diseases and the discovery of targeted candidates for new drugs in the future.”

 

Seung-man Han, CEO of Bertis, said, “In our attempts to overcome the limits of proteomics data interpretation by integrating bioinformatics and AI technologies with proteomics technology, the publication of a paper on DeepMRM holds great significance. As a leader in proteomics in Asia, Bertis will spare no efforts in developing innovative technologies that can fully unlock the potential of proteomics in R&D and actual clinical applications.”

 

Bertis is devoting itself to a project to eliminate limitations on interpreting and using proteomics data by incorporating proteomics technology accumulated over 10 years with bioinformatics and AI technologies. In particular, the company is making R&D efforts for the SAN (Spectrum is All you Need) model with a view to developing an innovative health information interpretation model which can produce health information from proteomic biometric information by automating the information processing and interpretation process.


go top