{"id":6175,"date":"2025-09-16T10:47:15","date_gmt":"2025-09-16T02:47:15","guid":{"rendered":"https:\/\/nullthought.net\/?p=6175"},"modified":"2025-09-16T10:47:17","modified_gmt":"2025-09-16T02:47:17","slug":"nucleobench%ef%bc%9a%e5%9f%ba%e4%ba%8e%e7%a5%9e%e7%bb%8f%e7%bd%91%e7%bb%9c%e7%9a%84%e6%a0%b8%e9%85%b8%e8%ae%be%e8%ae%a1%e7%ae%97%e6%b3%95%e5%a4%a7%e8%a7%84%e6%a8%a1%e5%9f%ba%e5%87%86","status":"publish","type":"post","link":"https:\/\/nullthought.net\/?p=6175","title":{"rendered":"NucleoBench\uff1a\u57fa\u4e8e\u795e\u7ecf\u7f51\u7edc\u7684\u6838\u9178\u8bbe\u8ba1\u7b97\u6cd5\u5927\u89c4\u6a21\u57fa\u51c6"},"content":{"rendered":"\n<p>\u8bba\u6587<strong><a href=\"https:\/\/www.biorxiv.org\/content\/10.1101\/2025.06.20.660785v3\" target=\"_blank\" rel=\"noreferrer noopener\">NucleoBench: A Large-Scale Benchmark of Neural Nucleic Acid Design Algorithms<\/a><\/strong>\u6784\u5efa\u4e86\u76ee\u524d\u89c4\u6a21\u6700\u5927\u7684\u6838\u9178\u8bbe\u8ba1\u57fa\u51c6 NucleoBench\u3002<\/p>\n\n\n\n<p>\u8bba\u6587\u4f5c\u8005\u4e3aJoel Shor\uff0cErik Strand\uff0cCory Y. McLean\uff0c\u6765\u81eaMove37 Labs\uff0c MIT Center for Bits and Atoms\u548cGoogle Research\u3002<\/p>\n\n\n\n<p>\u4e00\u3001\u7814\u7a76\u80cc\u666f\u4e0e\u52a8\u673a<br>\u6838\u9178\u5e8f\u5217\uff08DNA\/RNA\uff09\u5b9a\u5411\u8bbe\u8ba1\u662f\u836f\u7269\u7814\u53d1\u7684\u91cd\u8981\u74f6\u9888\uff1a\u4ec5\u4ec5 5\u2019 UTR \u7684\u641c\u7d22\u7a7a\u95f4\u5c31\u9ad8\u8fbe\u7ea6 2\u00d710<sup>120<\/sup>\uff0c\u7a77\u4e3e\u4e0d\u53ef\u884c\u3002\u5c3d\u7ba1\u8fd1\u5e74\u6765\u9488\u5bf9\u8f6c\u5f55\u56e0\u5b50\u7ed3\u5408\u3001\u67d3\u8272\u8d28\u53ef\u53ca\u6027\u4e0e\u57fa\u56e0\u8868\u8fbe\u7b49\u6027\u8d28\u7684\u9884\u6d4b\u6a21\u578b\u5feb\u901f\u8fdb\u6b65\uff0c\u4f46\u201c\u5982\u4f55\u4ece\u6a21\u578b\u4e2d\u53cd\u63a8\u9ad8\u6027\u80fd\u5e8f\u5217\u201d\u7684\u4f18\u5316\u7b97\u6cd5\u7f3a\u4e4f\u7cfb\u7edf\u57fa\u51c6\uff0c\u76f4\u63a5\u9650\u5236\u4e86\u4ece\u9ad8\u8d28\u91cf\u6a21\u578b\u4ea7\u51fa\u9ad8\u8d28\u91cf\u5206\u5b50\u7684\u80fd\u529b\u3002NucleoBench \u65e8\u5728\u586b\u8865\u8fd9\u4e00\u7a7a\u767d\uff0c\u9762\u5411 16 \u4e2a\u751f\u7269\u5b66\u4efb\u52a1\u30019 \u7c7b\u7ecf\u5178\u4e0e\u6df7\u5408\u201c\u8bbe\u8ba1\u5668\u201d\uff08\u4f18\u5316\u7b97\u6cd5\uff09\u5f00\u5c55\u5927\u89c4\u6a21\u5bf9\u6bd4\u5b9e\u9a8c\uff08>40 \u4e07\u6b21\uff09\uff0c\u7cfb\u7edf\u56de\u7b54\u8d85\u53c2\u6570\u3001\u521d\u59cb\u5e8f\u5217\u4e0e\u968f\u673a\u6027\u7684\u4f5c\u7528\uff0c\u5e76\u636e\u6b64\u63d0\u51fa\u65b0\u7b97\u6cd5 AdaBeam\u3002<\/p>\n\n\n\n<p>\u4e8c\u3001\u5de5\u4f5c\u4e0e\u8d21\u732e\u6982\u8ff0<br>1\uff09\u8986\u76d6\u201c\u957f\u5e8f\u5217\/\u5927\u6a21\u578b\u201d\u573a\u666f\uff1b2\uff09\u5bf9\u6807\u51c6\u4e0e\u65b0\u9896\u8bbe\u8ba1\u5668\u5728 16 \u4e2a\u4efb\u52a1\u4e0a\u8fdb\u884c 40 \u4e07+\u5b9e\u9a8c\uff1b3\uff09\u63d0\u4f9b\u5173\u4e8e\u8d77\u59cb\u8d85\u53c2\u6570\u5408\u7406\u533a\u95f4\u3001\u5bf9\u968f\u673a\u79cd\u5b50\/\u8d77\u59cb\u5e8f\u5217\u7684\u654f\u611f\u6027\u3001\u968f\u6a21\u578b\/\u5e8f\u5217\u957f\u5ea6\u6269\u5c55\u6027\u7684\u201c\u6570\u636e\u9a71\u52a8\u7b54\u6848\u201d\uff1b4\uff09\u57fa\u4e8e\u6d1e\u89c1\u63d0\u51fa AdaBeam\uff0c\u5728 16 \u4e2a\u4efb\u52a1\u4e2d\u6709 11 \u4e2a\u80dc\u51fa\uff0c\u5e76\u5728\u5927\u6a21\u578b\u957f\u5e8f\u5217\u4e0a\u5177\u66f4\u4f18\u6269\u5c55\u6027\u3002<a href=\"https:\/\/github.com\/move37-labs\/nucleobench\" target=\"_blank\" rel=\"noreferrer noopener\">\u4ee3\u7801\u5f00\u6e90<\/a>\u3002<\/p>\n\n\n\n<p>\u4e09\u3001\u57fa\u51c6\u4efb\u52a1\u4e0e\u6570\u636e\u8bbe\u5b9a<br>\u4efb\u52a1\u6765\u81ea\u56db\u4e2a\u7c7b\u522b\uff08\u8868 3\uff09\uff1a\uff081\uff09CRE\uff1a\u7ec6\u80de\u7c7b\u578b\u7279\u5f02\u7684\u987a\u5f0f\u8c03\u63a7\u6d3b\u6027\uff08Malinois \u6a21\u578b\uff0c3 \u4e2a\u4efb\u52a1\uff0c\u5e8f\u5217 200 bp\uff09\uff1b\uff082\uff09TF \u7ed3\u5408\uff1a\u8f6c\u5f55\u56e0\u5b50\u5728 3 kb \u533a\u6bb5\u7684\u7ed3\u5408\u6982\u7387\uff08BPNet-lite\uff0c11 \u4e2a\u4efb\u52a1\uff09\uff1b\uff083\uff09ATAC\uff1a\u67d3\u8272\u8d28\u53ef\u53ca\u6027\uff08BPNet-lite\uff0c1 \u4e2a\u4efb\u52a1\uff09\uff1b\uff084\uff09\u9009\u62e9\u6027\u57fa\u56e0\u8868\u8fbe\uff1aEnformer \u9884\u6d4b\u8868\u8fbe\u5e76\u505a\u808c\u8089\/\u795e\u7ecf\u9009\u62e9\u6027\u4f18\u5316\uff08\u8f93\u5165 200K bp\uff0c\u4ec5\u5141\u8bb8\u7f16\u8f91\u5176\u4e2d 256 \u4e2a\u78b1\u57fa\uff09\u3002\u5176\u4e2d BPNet-lite \u4efb\u52a1\u4f7f\u7528 ENCODE K562 \u6570\u636e\uff1bEnformer \u4efb\u52a1\u4e3a\u6587\u732e\u6240\u89c1\u9996\u6b21\u7528\u4e8e\u8bbe\u8ba1\uff0c\u4e14\u53ef\u7f16\u8f91\u4f4d\u70b9\u57fa\u4e8e Enformer \u9884\u6d4b\u7684 DNASE \u6d3b\u6027\u9009\u53d6\u3002<\/p>\n\n\n\n<p>\u56db\u3001\u8bbe\u8ba1\u5668\uff08\u4f18\u5316\u7b97\u6cd5\uff09\u8c31\u7cfb<br>\u57fa\u51c6\u5907\u6848\u8986\u76d6 9 \u4e2a\u4ee3\u8868\u6027\u201c\u8bbe\u8ba1\u5668\u201d\uff1aDirected Evolution\u3001Simulated Annealing\u3001AdaLead\u3001FastSeqProp\u3001Ledidi\uff0c\u4ee5\u53ca\u4f5c\u8005\u5f15\u5165\u5e76\u7edf\u4e00\u5b9e\u73b0\u7684 Ordered\/Unordered Beam\u3001Gradient Evo \u4e0e AdaBeam\uff08\u6df7\u5408\/\u6539\u8fdb\u65b9\u6cd5\uff09\u3002\u5176\u4e2d Ledidi\u3001FastSeqProp \u5c5e\u68af\u5ea6\u578b\uff1bGradient Evo \u4e0e AdaBeam \u4e3a\u6df7\u5408\/\u6539\u8fdb\u578b\u3002<\/p>\n\n\n\n<p>\u4e94\u3001\u5b9e\u9a8c\u8bbe\u7f6e\u4e0e\u8bc4\u6d4b\u6307\u6807<br>\u6bcf\u4e2a\u5b9e\u9a8c\u7531\u201c\u4efb\u52a1\u3001\u8bbe\u8ba1\u5668\u3001\u8bbe\u8ba1\u5668\u8d85\u53c2\u6570\u3001\u8d77\u59cb\u5e8f\u5217\u201d\u56db\u8981\u7d20\u786e\u5b9a\u3002Enformer \u4efb\u52a1\u5355\u6b21\u8fd0\u884c 12 \u5c0f\u65f6\uff0c\u5176\u4ed6\u4efb\u52a1 8 \u5c0f\u65f6\u6216\u6ee1\u8db3\u7b97\u6cd5\u7ec8\u6b62\u6761\u4ef6\uff1b\u5168\u90e8\u5728 CPU-only \u7684 n1-highmem-16\uff08Google Batch\uff09\u4e0a\u8fd0\u884c\u3002\u4e3b\u6307\u6807\u4e3a\u6700\u7ec8\u201c\u80fd\u91cf\/\u9002\u5e94\u5ea6\u201d\uff08\u7531\u4efb\u52a1\u6a21\u578b\u8bc4\u4f30\uff09\uff0c\u5e76\u8bb0\u5f55\u4f18\u5316\u8fc7\u7a0b\u7684\u80fd\u91cf\u66f2\u7ebf\u4e0e\u6bcf\u6b65\u8017\u65f6\uff0c\u7edf\u8ba1\u63a8\u65ad\u4f7f\u7528\u914d\u5bf9\u68c0\u9a8c\u4e0e 95% \u7f6e\u4fe1\u533a\u95f4\u3002<\/p>\n\n\n\n<p>\u516d\u3001\u8d77\u59cb\u5e8f\u5217\u4e0e\u968f\u673a\u6027\u7684\u7cfb\u7edf\u6027\u7814\u7a76<br>\u4e3a\u516c\u5e73\u6bd4\u8f83\u4e0e\u589e\u5f3a\u7edf\u8ba1\u529f\u6548\uff0c\u4f5c\u8005\u4e3a\u6bcf\u4e2a\u4efb\u52a1\u56fa\u5b9a 100 \u6761\u8d77\u59cb\u5e8f\u5217\uff08\u76f8\u540c\u8d77\u70b9\u7528\u4e8e\u6240\u6709\u8bbe\u8ba1\u5668\uff09\uff0c\u5e76\u5728\u6700\u4f73\u201c\u5f00\u7bb1\u5373\u7528\u201d\u8d85\u53c2\u6570\u4e0b\u5bf9\u6bcf\u7ec4\uff08\u4efb\u52a1\u00d7\u7b97\u6cd5\u00d7\u8d85\u53c2\u00d7\u8d77\u59cb\uff09\u91cd\u590d 5 \u4e2a\u968f\u673a\u79cd\u5b50\uff0c\u6784\u9020\u57fa\u4e8e\u79e9\u7684\u201c0 \u8d77\u70b9\u201d\u975e\u53c2\u6570\u5e8f\u5206\u6570\u6765\u8861\u91cf\u65b9\u5dee\uff08\u8d8a\u4f4e\u8d8a\u7a33\uff09\u3002\u540c\u65f6\u4f7f\u7528 Friedman + Nemenyi \u65b9\u6cd5\u5206\u6790\u201c\u56fa\u6709\u56f0\u96be\u8d77\u59cb\u5e8f\u5217\u201d\u7684\u5b58\u5728\u6027\u4e0e\u5206\u5e03\u3002<\/p>\n\n\n\n<p>\u4e03\u3001\u5173\u952e\u5de5\u7a0b\u4e0e\u65b9\u6cd5\u5b66\u521b\u65b0\uff1a\u5927\u6a21\u578b\u53ef\u53cd\u5411\u4f20\u64ad\u7684\u201c\u68af\u5ea6\u62fc\u63a5\u201d<br>Ledidi\/FastSeqProp \u7b49\u5e38\u7528\u201c\u68af\u5ea6\u63a9\u7801\u201d\u505a\u8d85\u957f\u5e8f\u5217\u7684\u5c40\u90e8\u53cd\u4f20\uff0c\u867d\u6b63\u786e\u4f46\u4e0d\u80fd\u964d\u4f4e\u5cf0\u503c\u663e\u5b58\uff0c\u5bfc\u81f4 Enformer\uff08\u7ea6 200K \u8f93\u5165\uff09\u4e0a\u6613 OOM\u3002\u4f5c\u8005\u63d0\u51fa\u201c\u68af\u5ea6\u62fc\u63a5\u201d\uff08Gradient Concatenation\uff09\uff1a\u628a\u9700\u8981\u6c42\u68af\u5ea6\u7684\u5207\u7247\u4e0e\u4e0d\u6c42\u68af\u5ea6\u7684\u5207\u7247\u5728\u5f20\u91cf\u7ef4\u5ea6\u4e0a\u62fc\u63a5\uff0c\u4ec5\u5bf9\u524d\u8005\u5efa\u56fe\u5e76\u53cd\u4f20\uff0c\u4ece\u800c\u663e\u8457\u964d\u4f4e\u53cd\u5411\u5f00\u9500\uff0c\u4f7f\u5f97 Enformer \u4efb\u52a1\u4e0a\u7684\u68af\u5ea6\/\u6df7\u5408\u6cd5\u6210\u4e3a\u53ef\u80fd\uff08\u800c Ledidi\/FastSeqProp \u4e0d\u80fd\uff09\u3002<\/p>\n\n\n\n<p>\u516b\u3001\u6539\u8fdb\u4e0e\u65b0\u7b97\u6cd5\uff1aAdaBeam \u4e0e Gradient Evo<br>1\uff09AdaBeam\uff1a\u53d7 Beam Search \u4e0e AdaLead \u542f\u53d1\uff0c\u628a\u6bcf\u8f6e\u4f18\u5316\u62c6\u4e3a\u201c\u9009\u4f4d\u7f6e\/\u505a\u7f16\u8f91\u201d\u4e24\u6b65\uff0c\u4f46\u5c06 AdaLead \u4e2d\u9690\u5f0f O(n)O(n)O(n) \u7684\u4f4d\u7f6e\u9009\u62e9\u66ff\u6362\u4e3a\u663e\u5f0f\u91c7\u6837\u7684 O(1)O(1)O(1)\uff0c\u5728 3 kb \u7ea7\u5e8f\u5217\u4e0a\u6b65\u8fdb\u901f\u5ea6\u7ea6\u4e3a AdaLead \u7684 2 \u500d\uff1b\u5e76\u7528\u76f4\u63a5\u53ef\u91c7\u6837\u7684\uff08\u4fee\u6b63\u540e\uff09\u4e8c\u9879\u5206\u5e03\u63a7\u5236\u6bcf\u8f6e\u7f16\u8f91\u6570\uff0c\u907f\u514d AdaLead \u7684\u62d2\u7edd\u91c7\u6837\u4f4e\u6548\u4e0e\u6f5c\u5728\u6b7b\u5faa\u73af\u3002<br>2\uff09Gradient Evo\uff1a\u5728 Directed Evolution \u6846\u67b6\u5185\u7528\u201c\u6cf0\u52d2\u8fd1\u4f3c\u7684 in-silico \u7a81\u53d8\u201d\uff08TISM\uff09\u6307\u5bfc\u7f16\u8f91\u4f4d\u7f6e\u9009\u62e9\uff0c\u628a\u6700\u5f31\u57fa\u7ebf\u63d0\u5347\u4e3a\u5f3a\u529b\u4f18\u5316\u5668\uff1b\u914d\u5bf9\u68c0\u9a8c\u663e\u793a\u201c\u4ec5\u5728\u9009\u70b9\u9636\u6bb5\u7528\u68af\u5ea6\u201d\u4e0e\u201c\u9009\u70b9+\u6362\u78b1\u57fa\u90fd\u7528\u68af\u5ea6\u201d\u65e0\u663e\u8457\u5dee\u5f02\uff0c\u63d0\u793a\u201c\u5148\u9009\u5bf9\u4f4d\u7f6e\u201d\u6bd4\u201c\u9009\u5bf9\u5177\u4f53\u66ff\u6362\u201d\u66f4\u5173\u952e\u3002<\/p>\n\n\n\n<p>\u4e5d\u3001\u4e3b\u8981\u7ed3\u679c\uff08\u6027\u80fd\uff09<br>\u8de8\u4efb\u52a1\u7684\u914d\u5bf9\u68c0\u9a8c\u663e\u793a\uff1aAdaBeam \u5e73\u5747\u6027\u80fd\u6700\u4f73\uff0c\u5e76\u4e0e Ledidi \u5e76\u5217\u4e3a\u6700\u5f3a\u4e24\u7c7b\uff1b\u5728 16 \u4e2a\u4efb\u52a1\u4e2d\uff0cAdaBeam \u4e8e 15 \u4e2a\u4efb\u52a1\u4e0a\u53d6\u5f97\u7b2c 1 \u6216\u7b2c 2\uff0cLedidi \u4e8e 14 \u4e2a\u4efb\u52a1\u4e0a\u53d6\u5f97\u7b2c 1 \u6216\u7b2c 2\u3002\u4efb\u52a1\u4e4b\u95f4\u96be\u5ea6\u5206\u5316\u663e\u8457\uff1aMalinois \u7c7b\u4efb\u52a1\u5bf9\u591a\u6570\u7b97\u6cd5\u201c\u53ef\u89e3\u201d\uff0c\u800c BPNet\/Enformer \u4f53\u73b0\u66f4\u5927\u533a\u5206\u5ea6\u3002<\/p>\n\n\n\n<p>\u5341\u3001\u7a33\u5b9a\u6027\u4e0e\u6536\u655b\u6027<br>\u968f\u673a\u79cd\u5b50\u5e26\u6765\u7684\u65b9\u5dee\u6574\u4f53\u4e0d\u5927\uff08\u6a21\u62df\u9000\u706b\u6700\u4f4e\uff0c\u4f46\u5176\u5e73\u5747\u6027\u80fd\u8f83\u5f31\uff09\uff1b\u8d77\u59cb\u5e8f\u5217\u5bf9\u6027\u80fd\u7684\u5f71\u54cd\u663e\u8457\uff0c\u4e14\u5b58\u5728\u201c\u56fa\u6709\u56f0\u96be\u8d77\u59cb\u5e8f\u5217\u201d\uff0c\u4e0d\u540c\u4efb\u52a1\u7684\u56f0\u96be\u5ea6\u5206\u5e03\u4e0d\u5747\uff1b\u6536\u655b\u901f\u5ea6\u4ee5\u975e\u53c2\u6570\u79e9\u5206\u6570\u6c47\u603b\uff0c\u7ed9\u51fa\u5404\u8bbe\u8ba1\u5668\u7684\u76f8\u5bf9\u5feb\u6162\u6392\u5e8f\uff0c\u7528\u4e8e\u5de5\u7a0b\u9009\u578b\u3002<\/p>\n\n\n\n<p>\u5341\u4e00\u3001AdaLead \u6d88\u878d\u4e0e\u65b9\u6cd5\u5b66\u6d1e\u89c1<br>\u7cfb\u7edf\u6d88\u878d\u8868\u660e\uff1a\u5173\u95ed\u91cd\u7ec4\uff08recombination\uff09\u4e0e\u9608\u503c\uff08thresholding\uff09\u540e\u6027\u80fd\u663e\u8457\u63d0\u5347\uff1b\u540c\u65f6\u4ee5\u663e\u5f0f\u5206\u5e03\u66ff\u6362\u62d2\u7edd\u91c7\u6837\u53ef\u628a\u9690\u5f0f O(n)O(n)O(n) \u8fc7\u7a0b\u5316\u4e3a O(1)O(1)O(1)\uff0c\u907f\u514d\u957f\u5e8f\u5217\u4e0b\u7684\u6548\u7387\u4e0e\u7a33\u5b9a\u6027\u95ee\u9898\u3002\u8fd9\u4e9b\u6d1e\u89c1\u5171\u540c\u50ac\u751f\u4e86 AdaBeam \u7684\u8bbe\u8ba1\u3002<\/p>\n\n\n\n<p>\u5341\u4e8c\u3001\u4e3a\u4f55\u201c\u68af\u5ea6\u62fc\u63a5\u201d\u5f88\u5173\u952e<br>\u5728 Enformer \u7b49\u201c\u524d\u5411\u5c1a\u53ef\u3001\u53cd\u5411\u6781\u91cd\u201d\u7684\u573a\u666f\u4e2d\uff0c\u4ec5\u5728\u9700\u8981\u7684\u5207\u7247\u4e0a\u5efa\u56fe\u53cd\u4f20\uff0c\u53ef\u628a\u5cf0\u503c\u663e\u5b58\u4ece\u201c\u5168\u5e8f\u5217\u63a9\u7801\u201d\u7ea7\u522b\u964d\u5230\u201c\u5c40\u90e8\u5207\u7247\u201d\u7ea7\u522b\uff0c\u8fd9\u662f\u8ba9\u68af\u5ea6\/\u6df7\u5408\u6cd5\u9996\u6b21\u771f\u6b63\u53ef\u7528\u4e8e 200K \u8f93\u5165\u7684\u5927\u6a21\u578b\u5e8f\u5217\u8bbe\u8ba1\u7684\u6838\u5fc3\u5de5\u7a0b\u6280\u5de7\u3002<\/p>\n\n\n\n<p>\u5341\u4e09\u3001\u5c40\u9650\u6027<br>\u7ed3\u8bba\u7684\u201c\u751f\u7269\u76f8\u5173\u6027\u201d\u53d6\u51b3\u4e8e\u4efb\u52a1\u6a21\u578b\u7684\u51c6\u786e\u6027\uff1b\u672c\u6587\u672a\u7cfb\u7edf\u6bd4\u8f83\u7b97\u6cd5\u5bf9\u751f\u7269\u5b66\u7ea6\u675f\/\u53ef\u884c\u6027\u7684\u7eb3\u5165\uff0c\u4e5f\u672a\u8bc4\u4f30\u751f\u6210\u5f0f\u6a21\u578b\u4e0e\u591a\u76ee\u6807\u6743\u8861\u7684\u5f71\u54cd\u3002<\/p>\n\n\n\n<p>\u5341\u56db\u3001\u672a\u6765\u5de5\u4f5c<br>\u8ba1\u5212\u6269\u5c55\u5230\u66f4\u591a\u4efb\u52a1\uff08\u542b\u5408\u6210\u57fa\u51c6\u3001\u957f\u7a0b\u89c4\u5212\u3001\u591a\u76ee\u6807\u4f18\u5316\uff09\u3001\u7814\u7a76\u968f\u5e8f\u5217\u957f\u5ea6\u4e0e\u6a21\u578b\u89c4\u6a21\u7684\u6269\u5c55\u6027\uff0c\u5f15\u5165\u9762\u5411\u4e0d\u786e\u5b9a\u6027\u7684\u4f18\u5316\u5668\u4e0e\u66f4\u5408\u7406\u7684\u63a2\u7d22\u2014\u5229\u7528\u5e73\u8861\u7b56\u7565\uff0c\u5e76\u5728\u4f18\u5316\u524d\u540e\u878d\u5165\u201c\u751f\u7269\u53ef\u884c\u6027\u7ea6\u675f\/\u7b5b\u9009\u201d\uff0c\u63d0\u5347\u6cbb\u7597\u5b66\u76f8\u5173\u6027\u3002<\/p>\n\n\n\n<p>\u5341\u4e94\u3001\u5bf9\u836f\u7269\u5f00\u53d1\u7684\u542f\u793a<br>\u6539\u8fdb\u7684\u5e8f\u5217\u8bbe\u8ba1\u53ef\u76f4\u63a5\u670d\u52a1\u4e8e\u591a\u4e2a\u5173\u952e\u5e94\u7528\uff1aCRISPR gRNA \u7279\u5f02\u6027\u4e0e\u8131\u9776\u4f18\u5316\u3001mRNA \u75ab\u82d7\u7ffb\u8bd1\u6548\u7387\u4e0e\u7a33\u5b9a\u6027\u63d0\u5347\u3001\u53cd\u4e49\u5be1\u6838\u82f7\u9178\uff08ASO\uff09\u9776\u5411\u4eb2\u548c\u4e0e\u514d\u75ab\u539f\u6027\u6298\u8877\u7b49\u3002<\/p>\n\n\n\n<p>\u5341\u516d\u3001\u5b9e\u8df5\u8005\u6307\u5357\uff08\u57fa\u4e8e\u672c\u6587\u53d1\u73b0\u7684\u5efa\u8bae\uff09<br>1\uff09\u9996\u9009\u8bbe\u8ba1\u5668\uff1a\u82e5\u5141\u8bb8\u4f7f\u7528\u68af\u5ea6\/\u6df7\u5408\u6cd5\uff0c\u4f18\u5148\u8003\u8651 AdaBeam \u4e0e Ledidi\uff1b\u5728\u5927\u6a21\u578b\/\u957f\u5e8f\u5217\u4e0b\uff0c\u52a1\u5fc5\u91c7\u7528\u201c\u68af\u5ea6\u62fc\u63a5\u201d\u7b49\u5185\u5b58\u4f18\u5316\u4ee5\u786e\u4fdd\u53ef\u8fd0\u884c\u6027\u3002<br>2\uff09\u8d85\u53c2\u6570\u542f\u53d1\u5f0f\uff1a\u8fc1\u79fb\u65e2\u6709\u57fa\u51c6\u7684\u4f18\u9009\u533a\u95f4\uff0c\u5e76\u968f\u5e8f\u5217\u957f\u5ea6\u9002\u5f53\u4e0b\u8c03\u7a81\u53d8\u7387\/\u6b65\u5e45\uff1b\u7528\u5c0f\u89c4\u6a21\u7f51\u683c\u6216\u8d1d\u53f6\u65af\u641c\u7d22\u5728\u7ed9\u5b9a\u4efb\u52a1\u5148\u7c97\u8c03\u3002<br>3\uff09\u8d77\u59cb\u5e8f\u5217\u63a7\u5236\uff1a\u7edf\u4e00\u4f7f\u7528\u540c\u4e00\u6279\u8d77\u70b9\u505a\u7b97\u6cd5\u6bd4\u8f83\uff1b\u5bf9\u751f\u4ea7\u5e94\u7528\uff0c\u5efa\u8bae\u5bf9\u82e5\u5e72\u201c\u4ee3\u8868\u6027\u8d77\u70b9\u201d\u914d\u5bf9\u8bc4\u4f30\uff0c\u4ee5\u89c4\u907f\u88ab\u201c\u56f0\u96be\u79cd\u5b50\u201d\u8bef\u5bfc\u7684\u98ce\u9669\u3002<br>4\uff09\u7edf\u8ba1\u8bc4\u4f30\uff1a\u5c3d\u91cf\u91c7\u7528\u914d\u5bf9\u8bbe\u7f6e\uff08\u56fa\u5b9a\u8d77\u70b9\/\u591a\u79cd\u5b50\uff09\uff0c\u7528\u975e\u53c2\u6570\u79e9\u5206\u6570\u6c47\u603b\u8de8\u4efb\u52a1\u7a33\u5065\u6027\u4e0e\u6536\u655b\u6027\u3002<br>5\uff09\u7b97\u6cd5\u6539\u9020\u8981\u70b9\uff1a\u5728\u8fdb\u5316\/\u7fa4\u4f53\u7c7b\u65b9\u6cd5\u4e2d\uff0c\u4f18\u5148\u628a\u201c\u7f16\u8f91\u4f4d\u7f6e\u9009\u62e9\u201d\u505a\u5f97\u66f4\u806a\u660e\uff08\u5982\u7528\u68af\u5ea6\/TISM \u6307\u5bfc\uff09\uff0c\u5176\u6536\u76ca\u53ef\u80fd\u5927\u4e8e\u5bf9\u5177\u4f53\u66ff\u6362\u7684\u7cbe\u96d5\u7ec6\u7422\u3002<\/p>\n\n\n\n<p>\u5341\u4e03\u3001\u603b\u7ed3<br>NucleoBench \u901a\u8fc7\u7edf\u4e00\u4efb\u52a1\u4e0e\u8d77\u70b9\u3001\u653e\u5927\u91cf\u7ea7\u5e76\u5f15\u5165\u5173\u952e\u5de5\u7a0b\u6280\u5de7\uff0c\u7ed9\u51fa\u4e86\u5f53\u524d\u6838\u9178\u5e8f\u5217\u8bbe\u8ba1\u5668\u5728\u201c\u771f\u5b9e\u96be\u5ea6\u201d\u573a\u666f\u4e0b\u7684\u6e05\u6670\u753b\u50cf\uff1aAdaBeam \u4e0e Ledidi \u9886\u8dd1\uff0c\u8d77\u59cb\u5e8f\u5217\u5f71\u54cd\u4e0d\u53ef\u5ffd\u89c6\uff0c\u6536\u655b\u4e0e\u7a33\u5b9a\u6027\u53ef\u91cf\u5316\u6392\u5e8f\uff1b\u201c\u68af\u5ea6\u62fc\u63a5\u201d\u628a Enformer \u7ea7\u4efb\u52a1\u4ece\u201c\u4e0d\u53ef\u505a\u201d\u53d8\u4e3a\u201c\u53ef\u505a\u201d\u3002\u4f5c\u4e3a\u5f00\u653e\u57fa\u51c6\u4e0e\u53c2\u8003\u5b9e\u73b0\uff0c\u5b83\u4e3a\u540e\u7eed\u5728\u201c\u6a21\u578b\u4e0d\u786e\u5b9a\u6027\u3001\u53ef\u884c\u6027\u7ea6\u675f\u4e0e\u751f\u6210\u6a21\u578b\u201d\u7684\u66f4\u5168\u9762\u6bd4\u8f83\u5960\u5b9a\u4e86\u575a\u5b9e\u57fa\u7ebf\u3002<\/p>\n\n\n\n<hr class=\"wp-block-separator has-alpha-channel-opacity\"\/>\n\n\n\n<p>NucleoBench on GitHub\uff1a<a href=\"https:\/\/github.com\/move37-labs\/nucleobench\" target=\"_blank\" rel=\"noreferrer noopener\">https:\/\/github.com\/move37-labs\/nucleobench<\/a><\/p>\n","protected":false},"excerpt":{"rendered":"<p>\u8bba\u6587NucleoBench: A Large-Scale Benchmark of Neural Nuclei [&hellip;]<\/p>\n","protected":false},"author":1,"featured_media":0,"comment_status":"open","ping_status":"open","sticky":false,"template":"","format":"standard","meta":{"site-sidebar-layout":"default","site-content-layout":"","ast-site-content-layout":"default","site-content-style":"default","site-sidebar-style":"default","ast-global-header-display":"","ast-banner-title-visibility":"","ast-main-header-display":"","ast-hfb-above-header-display":"","ast-hfb-below-header-display":"","ast-hfb-mobile-header-display":"","site-post-title":"","ast-breadcrumbs-content":"","ast-featured-img":"","footer-sml-layout":"","ast-disable-related-posts":"","theme-transparent-header-meta":"","adv-header-id-meta":"","stick-header-meta":"","header-above-stick-meta":"","header-main-stick-meta":"","header-below-stick-meta":"","astra-migrate-meta-layouts":"set","ast-page-background-enabled":"default","ast-page-background-meta":{"desktop":{"background-color":"","background-image":"","background-repeat":"repeat","background-position":"center center","background-size":"auto","background-attachment":"scroll","background-type":"","background-media":"","overlay-type":"","overlay-color":"","overlay-opacity":"","overlay-gradient":""},"tablet":{"background-color":"","background-image":"","background-repeat":"repeat","background-position":"center center","background-size":"auto","background-attachment":"scroll","background-type":"","background-media":"","overlay-type":"","overlay-color":"","overlay-opacity":"","overlay-gradient":""},"mobile":{"background-color":"","background-image":"","background-repeat":"repeat","background-position":"center center","background-size":"auto","background-attachment":"scroll","background-type":"","background-media":"","overlay-type":"","overlay-color":"","overlay-opacity":"","overlay-gradient":""}},"ast-content-background-meta":{"desktop":{"background-color":"var(--ast-global-color-5)","background-image":"","background-repeat":"repeat","background-position":"center center","background-size":"auto","background-attachment":"scroll","background-type":"","background-media":"","overlay-type":"","overlay-color":"","overlay-opacity":"","overlay-gradient":""},"tablet":{"background-color":"var(--ast-global-color-5)","background-image":"","background-repeat":"repeat","background-position":"center center","background-size":"auto","background-attachment":"scroll","background-type":"","background-media":"","overlay-type":"","overlay-color":"","overlay-opacity":"","overlay-gradient":""},"mobile":{"background-color":"var(--ast-global-color-5)","background-image":"","background-repeat":"repeat","background-position":"center center","background-size":"auto","background-attachment":"scroll","background-type":"","background-media":"","overlay-type":"","overlay-color":"","overlay-opacity":"","overlay-gradient":""}},"footnotes":""},"categories":[8,36],"tags":[39,45,66],"class_list":["post-6175","post","type-post","status-publish","format-standard","hentry","category-tech","category-36","tag-ai","tag-medicine","tag-google"],"rttpg_featured_image_url":null,"rttpg_author":{"display_name":"NullThought","author_link":"https:\/\/nullthought.net\/?author=1"},"rttpg_comment":0,"rttpg_category":"<a href=\"https:\/\/nullthought.net\/?cat=8\" rel=\"category\">Tech<\/a> <a href=\"https:\/\/nullthought.net\/?cat=36\" rel=\"category\">\u79d1\u5b66<\/a>","rttpg_excerpt":"\u8bba\u6587NucleoBench: A Large-Scale Benchmark of Neural Nuclei&hellip;","_links":{"self":[{"href":"https:\/\/nullthought.net\/index.php?rest_route=\/wp\/v2\/posts\/6175","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/nullthought.net\/index.php?rest_route=\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/nullthought.net\/index.php?rest_route=\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/nullthought.net\/index.php?rest_route=\/wp\/v2\/users\/1"}],"replies":[{"embeddable":true,"href":"https:\/\/nullthought.net\/index.php?rest_route=%2Fwp%2Fv2%2Fcomments&post=6175"}],"version-history":[{"count":1,"href":"https:\/\/nullthought.net\/index.php?rest_route=\/wp\/v2\/posts\/6175\/revisions"}],"predecessor-version":[{"id":6176,"href":"https:\/\/nullthought.net\/index.php?rest_route=\/wp\/v2\/posts\/6175\/revisions\/6176"}],"wp:attachment":[{"href":"https:\/\/nullthought.net\/index.php?rest_route=%2Fwp%2Fv2%2Fmedia&parent=6175"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/nullthought.net\/index.php?rest_route=%2Fwp%2Fv2%2Fcategories&post=6175"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/nullthought.net\/index.php?rest_route=%2Fwp%2Fv2%2Ftags&post=6175"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}