# 마우스 효과 팅커벨
본문 바로가기 메뉴 바로가기

벼랑끝 바이오 매드 컴퓨터 사이언티스트

프로필사진
  • 글쓰기
  • 관리
  • 태그
  • 방명록
  • RSS

벼랑끝 바이오 매드 컴퓨터 사이언티스트

검색하기 폼
  • 분류 전체보기 (173)
    • Background (64)
      • Biology (9)
      • Math (42)
      • Statistics (8)
      • English (5)
    • Computer (77)
      • 백준 풀이 (7)
      • 이게 왜 안되지? (27)
      • 코딩 개꿀팁 (28)
      • 3B1B 따라잡기 with ManimLibrary (15)
    • Paper (10)
      • Bioinformatics (6)
      • CS (4)
    • Daily (20)
      • review (6)
      • etc (14)
  • 방명록

전체 글 (173)
[이왜안?] AttABseq code error

Instead of running the main.py executable file located in the root of the AttAbSeq GitHub repository, you must run the main_AB___.py file located in the crossvalid/script/ directory.On line 10 of the script, the import statement from model import * must be changed. It should be modified to import from the corresponding model_AB___.py file, which is also located inside the crossvalid/script/ dire..

Computer/이게 왜 안되지? 2025. 9. 12. 15:53
챗지피티와의 아이디어 대화(1) - 현실적으로 FR과 CDR Codesign하는 생성 모델이 가능할까?

내 질문:실적으로 FR과 CDR Codesign하는 생성모델이 가능할것같아? 내가보기엔 데이터가 너무 적어서 학습하기 너무 어려울 것 같고, FR생성하는 모델 따로, CDR을 생성하는 모델을 따로 이용해서 두 모델이 생성한 구조를 Grafting하는 방법을 고안하고싶은데 이건 어떨까? ChatGPT의 대답:좋은 아이디어예요. 결론부터 말하면,FR+CDR 동시(co-design) 생성은 원리상 가능하지만, 데이터 희소성/라벨 부정확성 때문에 “바로 실전”은 어렵습니다.FR 모델과 CDR 모델을 분리해 각각의 강점을 살린 뒤, 가상 접목(grafting)과 강력한 사후-필터링으로 합치는 전략은 현실적이고 효과적인 트레이드오프입니다.더 나아가, 분리-생성+가교(graft) 파이프라인에 **약한 결합(soft ..

Daily/etc 2025. 9. 7. 00:10
항체 생성 모델 구축을 위한 배경지식

IMGT numbering system항체의 IMGT 구조는 다음과 같이 되어있음 (아래 그림 참고)A - B - BC Loop(CDR1)- C - C'- C'C'' Loop(CDR2) - C''- D - E - F - FG Loop(CDR3) - G Figure 1에 나타난 서로다른 구조들은 다음을 의미한다.초록색 선 : hydrogen bond를 형성하는 부분굵은 빨간색 아미노산 : Conserved AA Residues하늘색 배경색 : Hydrophobic AA Residue들이 50%이상 발견되는 부분들네모 : IMGT anchor (구조 경계)빗금 동그라미 : GapIMGT numbering system이 적용된 Ab들은 다음과 같은 특징들이 highly conserved 되어있다.1st-CYS..

Background/Biology 2025. 9. 2. 22:45
인공지능 기반 항체 공학 분야 논문 리뷰어 아카이브

논문 제출할 때 리뷰어 지정을 요구하는 저널들이 많아서 논문 읽으면서 관련분야 전문가라고 생각되는 분들, 논문 잘쓴다고 생각하는 분들은 저장해두는 편이 좋겠다고 생각이 들어서 작성하는 공간. Antigen-Antibody Interaction (AAI) predictionDeepInterAware : Yuhang Xia, Zhiwei Wang, Feng Huang, Zhankun Xiong, Yongkang Wang, Minyao Qiu, and Wen Zhang* Antibody Generation ModelArtificial intelligence-driven computational methods for antibody design and optimization : 김호민 교수님, 차미영 교수님 ..

Daily/etc 2025. 8. 26. 15:01
DeepInterAware 논문 리뷰 (2) 실험 및 결과

https://biomadscientist.tistory.com/185 DeepInterAware 논문 리뷰 (1) 모델 구조https://pmc.ncbi.nlm.nih.gov/articles/PMC11967782/ DeepInterAware: Deep Interaction Interface‐Aware Network for Improving Antigen‐Antibody Interaction Prediction from SequenceAbstract Identifying interactions between candidate antibodies and target antbiomadscientist.tistory.com이 논문 배울점이 많다. 모델 설명도 구체적이고 아이디어도 뭐 엄청나게 신박하지는 않지만..

Paper/Bioinformatics 2025. 8. 26. 13:46
DeepInterAware 논문 리뷰 (1) 모델 구조

https://pmc.ncbi.nlm.nih.gov/articles/PMC11967782/ DeepInterAware: Deep Interaction Interface‐Aware Network for Improving Antigen‐Antibody Interaction Prediction from SequenceAbstract Identifying interactions between candidate antibodies and target antigens is a key step in developing effective human therapeutics. The antigen–antibody interaction (AAI) occurs at the structural level, but the lim..

Paper/Bioinformatics 2025. 8. 25. 00:11
AttABseq 논문 리뷰

https://academic.oup.com/bib/article/25/4/bbae304/7705533서열 기반의 Antigen-Antibody Affinity change prediction 모델 Introduction매번 나오는 똑같은 Ab 이야기, 항체가 therapeutics에 효율적이다...specificity가 좋고 toxicity가 낮고 strong한 효과를 보인다...그러나 이런 Ab를 찾는데 기존 wet-lab 기반으로는 많은 시간과 돈을 들여야하는 단점이 있다.찾아야하는 sequence space가 너무 넓어서 wet-lab기반으로는 한계가 있다. 그러니 computational method가 필요하고 실제로 ML기반의 method들이 유망한 결과를 보였다 등 ..항체 최적화는 solu..

Paper/Bioinformatics 2025. 8. 22. 18:47
AbSet 논문리뷰

https://pubs.acs.org/doi/10.1021/acs.jcim.5c00410항체-단백질 복합체 데이터셋 생산에 대한 논문을 작성중이라 관련 논문을 찾아 읽어볼 필요가 있어서 정리해본다. 아마 이 논문이 우리가 진행하려던 아이디어와 상당부분 유사하기에 번역하듯 조금 꼼꼼히 읽으려한다. 1. Introduction1. Antibody 구조설명 VL, VH, Fab, Fc, Fv, CDRs (1~3) 2. Antibody의 CDR을 computational method를 통해서 optimization 해온 역사, 특히 AI 기반이면서 structure-based model들이 가장 유망한 결과를 보임 그러나 데이터의 부재를 문제점으로 지적함3. 현존하는 DB들의 문제점을 언급PDB에 antibod..

Paper/Bioinformatics 2025. 8. 19. 20:26
[이왜안?] ImportError: cannot import name 'LRScheduler' from 'torch.optim.lr_scheduler'

[원인]torch 버전과 torch를 이용한 training을 최적화해주는 lightening 라이브러리 사이의 업데이트 사이에서 변경점이 서로 충돌하는 것이 원인이라고 생각된다. torch의 lr_scheduler.py파일은 버전에 따라서 class 이름이 LRScheduler인 것도 있지만 _LRScheduler로 언더바가 붙은 버전도 배포된 것으로 생각된다. [해결법].conda/envs/{ENV_NAME}/lib/python3.8/site-packages/lightning_fabric/utilities/types.py로 들어가서 lightening의 LRScheduler import 코드에 언더바를 붙여서 저장해주면 된다.

Computer/이게 왜 안되지? 2025. 8. 7. 19:18
[이왜안?] DiffAb 코드구현 에러발생 ValueError: invalid literal for int() with base 10: 'X'

https://github.com/luost26/diffab/issues/21#issuecomment-2350808489 在运行sabdab.py代码,发生报错 · Issue #21 · luost26/diffab在运行sabdab.py代码,发生报错,请问该如何解决呀 第一个错误:报ValueError: invalid literal for int() with base 10: 'V'异常, 第二个错误如图,请教这两个问题都是怎么合理解决的呀github.comDiffAb 학습을 위해서 train 코드를 돌려보았는데 전처리과정에서 다음과 같은 에러가 나온다.File "train.py", line 55, in train_dataset = get_dataset(config.dataset.train) File "/data..

Computer/이게 왜 안되지? 2025. 8. 5. 18:52
이전 1 2 3 4 ··· 18 다음
이전 다음
공지사항
최근에 올라온 글
최근에 달린 댓글
Total
Today
Yesterday
링크
  • 벼랑끝 깃허브
  • 벼랑끝 강아지
TAG
  • elementry matrix
  • 인공지능
  • 오일석기계학습
  • 최대우도추정
  • 항원항체결합예측모델
  • 백준
  • antigen antibody interaction prediction
  • 선형대수
  • 이왜안
  • nanobody
  • marginal likelihood
  • 기계학습
  • Matrix algebra
  • manimtutorial
  • 베이즈정리
  • eigenvector
  • 논문리뷰
  • ai신약개발
  • manim library
  • eigenvalue
  • MLE
  • 나노바디
  • MorganCircularfingerprint
  • 3b1b
  • 파이썬
  • MatrixAlgebra
  • manim
  • Manimlibrary
  • dataloader
  • 3B1B따라잡기
more
«   2025/09   »
일 월 화 수 목 금 토
1 2 3 4 5 6
7 8 9 10 11 12 13
14 15 16 17 18 19 20
21 22 23 24 25 26 27
28 29 30
글 보관함

Blog is powered by Tistory / Designed by Tistory

티스토리툴바