소재 AI 공유함
Menu
← AI in materials

Materials AI · written by AI

AI Large-Scale Screening of Human Protein Scaffolds for mRNA Delivery Candidates, Performance Improved by TRIM25 Engineering

AI editorPublished 2026.10.31

The Nature Communications paper used an AI-based platform called DeepDelivery to screen human proteins at large scale for scaffold candidates for mRNA delivery. Proteome-scale screening proposed 512 candidates, 15 were selected, and 12 acted as actual scaffold components. Interpretive analysis identified TRIM domain features, and TRIM25 domain engineering improved delivery performance. [1]

1/ AI utilization method: DeepDelivery-based AI screening and interpretive analysis
2/ Applied task: Large-scale screening of scaffold candidates for mRNA delivery among human proteins
3/ Reported results: 512 candidates were proposed, 12 of 15 functioned as actual scaffold components, and delivery performance was improved by TRIM25 engineering
4/ This study used an AI-based platform called DeepDelivery to identify, at large scale, scaffold candidates among human proteins for use in mRNA delivery.
5/ Proteome-scale screening proposed 512 candidates, and when 15 were selected and tested, 12 functioned as actual scaffold components.
6/ In addition, interpretive analysis identified features related to the TRIM domain, and based on this, TRIM25 domain engineering was performed, leading to improved delivery performance.
7/ From a materials and chemistry perspective, this is an example in which AI-based screening and interpretable feature extraction were used together in the design of protein-based delivery carriers.
8/ However, these results were obtained for a specific protein set and mRNA delivery task, so they cannot be directly generalized to other protein families or other delivery conditions.
9/ Although experimental validation was included, there are still limitations in the number of candidates and the validation scope. [1]

How to apply this

Use AI to prioritize delivery carrier candidates in your own protein or polymer library, then experimentally validate a small number of candidates.

  1. First, organize the candidate sequences or structures into a standardized format and collect experimental values related to delivery performance.
  2. Next, score the candidates with an AI model and select the top candidates together with control groups.
  3. Evaluate the selected candidates in an actual delivery assay, and if necessary redesign them at the domain level.
  4. Reconfirm the candidates that performed well through a separate validation set and repeated experiments.
Data needed
A candidate sequence or structure library, information on the delivery cargo, a cellular or in vitro delivery evaluation system, and comparator control candidates are needed.
Checking the result
Check whether the AI-selected candidates function at a higher rate than random selection in experiments, and whether the same trend appears in an independent test set or additional replicate experiments.
What can go wrong
Predictions can easily fail for sequence ranges or delivery conditions different from the training data, and interpreted features should not be taken to be causal rules outright.

이 가이드는 원문의 방법을 바탕으로 AI가 제안한 적용 방향입니다. 원문에 적힌 내용이 아니며 같은 결과가 재현된다는 보장도 아닙니다.

Related concepts and practice

이 이슈를 읽는 데 필요한 개념입니다. 용어마다 Colab에서 바로 돌려볼 수 있는 실습 노트북이 있습니다.

Sources

  1. AI-driven discovery and engineering of human-derived protein scaffolds for mRNA delivery | Nature Communications

웹 자료를 바탕으로 AI가 자동 요약한 글입니다. 연구의 세부 조건과 결과는 연결된 원문을 확인해주세요.