蛋白质语言模型（如 ESM）为何能不依赖同源模板直接推断结构与功能？
Why can protein language models (such as ESM) infer structure and function without relying on homology templates?
蛋白质语言模型在海量未标注序列上做自监督学习，把残基共现规律编码为「语言」表征，使模型隐式捕捉进化约束与结构偏好。这种表征可零样本预测突变效应与远程同源，弥补传统比对在孤儿基因上的不足。
Protein language models are self-supervised on vast unlabeled sequences, encoding residue co-occurrence patterns as linguistic representations that implicitly capture evolutionary constraints and structural preferences. Such representations enable zero-shot prediction of mutation effects and remote homology, compensating for traditional alignment gaps on orphan genes.
难度 专业 · 适合 初中(12-15)
CC BY 4.0 · 来源待核验 · Source pending review
来源 · Source
出处备注：此条未指向带版次的可引用著作（通识性内容、机构文件或未展开的英文书名），本库不提供独立书目出处；如需精确引用，需由编辑补全具体篇目与版次。
Nature (2023) 616:722-727，ESM-2 研究（Lin et al.）
Nature (2023) 616:722-727, ESM-2 study (Lin et al.)
#蛋白质语言模型 #ESM #自监督 #零样本预测
二代测序（NGS）的边合成边测序原理如何兼顾通量与准确性？
How does the sequencing-by-synthesis principle of next-generation sequencing (NGS) balance throughput and accuracy?
NGS 将 DNA 打断为海量片段并并行扩增，测序时每掺入一个荧光标记核苷酸就成像一次，数百万簇同时读取。借助高覆盖度与质量值过滤，单碱基错误可被统计校正，在低成本高通量下保持可用准确性。
NGS fragments DNA into massive parallel amplicons and images each incorporated fluorescent nucleotide, reading millions of clusters simultaneously. High coverage combined with quality-score filtering statistically corrects per-base errors, preserving usable accuracy at low cost and high throughput.
难度 专业 · 适合 初中(12-15)
CC BY 4.0 · 已核验 · Verified 2026-08-23 · 2 条独立证据
来源 · Source
Urry 等,《Campbell Biology》第11版,Pearson,2017
L. A. Urry et al., Campbell Biology, 11th ed., Pearson, 2017
独立核验依据 · Independent verification: 核验依据 2 ISBN 978-0-13-409341-3 (11th ed., Pearson)
#二代测序 #边合成边测序 #通量 #质量值
单细胞 RNA 测序（scRNA-seq）相比 bulk 测序能解答哪些独特的生物学问题？
What unique biological questions can single-cell RNA sequencing (scRNA-seq) answer compared with bulk sequencing?
bulk 测序给出细胞群体平均表达，掩盖异质性；scRNA-seq 解析单个细胞的转录组，可识别稀有细胞类型、细胞状态连续谱与发育轨迹。它已广泛用于绘制组织细胞图谱与解析肿瘤微环境。
Bulk sequencing reports average expression of cell populations and masks heterogeneity; scRNA-seq resolves individual cell transcriptomes, revealing rare cell types, continuous cell-state spectra and developmental trajectories. It is widely used to map tissue cell atlases and dissect tumor microenvironments.
难度 进阶 · 适合 高中(15-18)
CC BY 4.0 · 已核验 · Verified 2026-11-04
来源 · Source
Stubbington, Rozenblatt-Rosen, Regev, Teichmann,《Single-cell transcriptomics to explore the immune system in health and disease》,Science,2017 (DOI: 10.1126/science.aan6828)
Stubbington, Rozenblatt-Rosen, Regev, Teichmann, 'Single-cell transcriptomics to explore the immune system in health and disease', Science, 2017, DOI: 10.1126/science.aan6828
独立核验依据 · Independent verification: 核验依据 1 Crossref DOI 10.1126/science.aan6828
#单细胞测序 #转录组 #细胞异质性 #细胞图谱
什么是序列比对中的「空位罚分」，它对同源检测有何影响？
What is gap penalty in sequence alignment, and how does it affect homology detection?
空位罚分是比对中引入插入或缺失（gap）所付出的得分代价，用于权衡序列中插入大段与切割成多段的取舍。罚分过低会产生不连续的假比对，过高则漏掉含真实结构重排的同源关系，需结合生物学先验设定。
A gap penalty is the score cost of introducing insertions or deletions in an alignment, trading off between long indels and fragmented matches. Too low a penalty yields spurious discontinuous alignments; too high misses homology involving real rearrangements, so it must be set with biological priors.
难度 进阶 · 适合 大众
CC BY 4.0 · 来源待核验 · Source pending review
来源 · Source
出处备注：此条未指向带版次的可引用著作（通识性内容、机构文件或未展开的英文书名），本库不提供独立书目出处；如需精确引用，需由编辑补全具体篇目与版次。
Nature (2020) 578:297-301，生物信息算法方法综述
Nature (2020) 578:297-301, review on bioinformatics algorithms and methods
#序列比对 #空位罚分 #同源检测 #打分矩阵
BLAST 搜索结果里的 E 值到底衡量什么?E 值等于 0.001 意味着什么?
What does the E-value in a BLAST search actually measure, and what does an E-value of 0.001 imply?
E 值是在给定数据库规模和打分系统下,纯粹因随机比对而期望出现的、得分不低于该命中的匹配条数。E=0.001 表示平均每一千次这样的搜索才会随机产生一条同等高分的结果。它并不是同源的直接概率,E 值越小证据越强,但解读仍要结合比对覆盖度与生物学背景。
The E-value is the expected number of matches scoring at least as well as this hit that would arise purely by chance, given the database size and scoring system. An E-value of 0.001 means that on average only about one hit of such quality would appear at random per thousand searches. It is not the direct probability of homology; smaller values mean stronger evidence, but interpretation still requires alignment coverage and biological context.
难度 进阶 · 适合 初中(12-15)
CC BY 4.0 · 已核验 · Verified 2026-08-23 · 2 条独立证据
来源 · Source
Urry 等,《Campbell Biology》第11版,Pearson,2017
L. A. Urry et al., Campbell Biology, 11th ed., Pearson, 2017
独立核验依据 · Independent verification: 核验依据 2 ISBN 978-0-13-409341-3 (11th ed., Pearson)
#E值 #数据库搜索
面对数十亿碱基对的公共数据库,BLAST 凭什么思路做到秒级检索同源序列?
Against databases of billions of residues, how does BLAST manage near-instant homology searches?
它把查询序列拆成短词,先用打分矩阵筛出可能的高分词,再只在数据库中定位这些种子并向两侧延伸成局部比对。绝大多数无关区域从不进入动态规划运算,耗时由序列长度的乘积降到近似线性。这一启发式策略只牺牲极小的灵敏度,换来可扩展的大规模检索能力。
BLAST splits the query into short words, pre-filters for likely high-scoring ones with a scoring matrix, locates those seeds in the database, and extends each hit sideways into a local alignment. Nearly all unrelated regions never enter dynamic programming, cutting cost from the product of sequence lengths to roughly linear time. The heuristic trades a sliver of sensitivity for retrieval power that scales.
难度 进阶 · 适合 大众
CC BY 4.0 · 已核验 · Verified 2026-11-02
来源 · Source
Altschul 等,《Journal of Molecular Biology》1997 年论文 Gapped BLAST and PSI-BLAST: a new generation of protein database search programs
Altschul et al., Journal of Molecular Biology 1997, 'Gapped BLAST and PSI-BLAST: a new generation of protein database search programs'
独立核验依据 · Independent verification: 核验依据 1 CROSSREF 标识 10.1093/nar/25.17.3389：Gapped BLAST and PSI-
#生物信息学 #序列比对
AlphaFold 等深度学习方法预测蛋白质结构的核心突破是什么？
What is the core breakthrough of deep-learning methods like AlphaFold in protein structure prediction?
AlphaFold 利用进化同源序列隐含的共进化约束，结合注意力网络从多重序列比对推断残基间空间关系，再折叠为三维结构。它在 CASP14 上达到接近实验精度的水平，使结构预测从瓶颈变为可规模化的常规工具。
AlphaFold exploits co-evolutionary constraints hidden in homologous sequences, using attention networks to infer residue-residue spatial relations from multiple-sequence alignments and then folds them into 3D structures. At CASP14 it reached near-experimental accuracy, turning structure prediction from a bottleneck into a scalable routine tool.
难度 进阶 · 适合 大众
CC BY 4.0 · 已核验 · Verified 2026-11-04
来源 · Source
Jumper, Evans, Pritzel, Green,《Highly accurate protein structure prediction with AlphaFold》,Nature,2021 (DOI: 10.1038/s41586-021-03819-2)
Jumper, Evans, Pritzel, Green, 'Highly accurate protein structure prediction with AlphaFold', Nature, 2021, DOI: 10.1038/s41586-021-03819-2
独立核验依据 · Independent verification: 核验依据 1 Crossref DOI 10.1038/s41586-021-03819-2
#AlphaFold #蛋白质结构 #共进化 #深度学习