nūhou

Hāʻawi ʻia ka Lasker Basic Medical Research Award i kēia makahiki iā Demis Hassabis a me John Jumper no kā lākou hāʻawi ʻana i ka hana ʻana o ka AlphaFold artificial intelligence system e wānana ana i ke ʻano ʻekolu-dimensional o nā protein e pili ana i ke kaʻina kauoha mua o nā waikawa amino.

ʻO kā lākou hopena e hoʻonā i kahi pilikia i hoʻonāukiuki lōʻihi i ke kaiāulu ʻepekema a wehe i ka puka no ka wikiwiki ʻana i ka noiʻi ma waena o ke kahua biomedical. He koʻikoʻi ko nā polokina i ka ulu ʻana o ka maʻi: i ka maʻi o Alzheimer, pelu lākou a hui pū; I ka maʻi kanesa, nalowale kā lākou hana hoʻoponopono; I loko o ka inborn metabolic disorders, ua dysfunctional; I loko o ka cystic fibrosis, hele lākou i kahi kūpono ʻole o ke kelepona. He liʻiliʻi wale nō kēia o nā ʻano hana e hoʻoulu ai i ka maʻi. Hiki i nā ʻano hoʻohālike kikoʻī o ka protein ke hāʻawi i nā hoʻonohonoho atomic, e hoʻokele i ka hoʻolālā a i ʻole ke koho ʻana i nā molekele pili kiʻekiʻe, a me ka wikiwiki i ka ʻike ʻana i ka lāʻau.

Hoʻoholo maʻamau ʻia nā ʻōnaehana protein e ka X-ray crystallography, nuclear magnetic resonance a me cryo-electron microscopy. He pipiʻi kēia mau ʻano hana a hoʻopau manawa. Loaʻa kēia i nā ʻikepili hoʻolālā protein 3D i loaʻa me kahi o 200,000 wale nō ka ʻikepili hoʻolālā, ʻoiai ua hoʻopuka ka ʻenehana sequencing DNA ma mua o 8 miliona mau kaʻina protein. I nā makahiki 1960, ʻo Anfinsen et al. Ua ʻike ʻo ia e hiki i ke kaʻina 1D o nā ʻakika amino ke hoʻopololei a hoʻopaʻa pinepine ʻia i loko o kahi hoʻohālikelike ʻekolu-dimensional (Figure 1A), a hiki i nā "chaperones" molekala ke hoʻoikaika a hoʻomaʻamaʻa i kēia kaʻina hana. Ke alakaʻi nei kēia mau ʻike i kahi luʻi 60 makahiki i ka biology molecular: wānana i ke ʻano 3D o nā protein mai ke kaʻina 1D o nā waikawa amino. Me ka holomua o ka Human Genome Project, ua hoʻomaikaʻi maikaʻi ʻia ko mākou hiki ke loaʻa i nā ʻāpana amino acid 1D, a ua ʻoi aku ka wikiwiki o kēia pilikia.

He paʻakikī ka wānana ʻana i nā ʻano protein no kekahi mau kumu. ʻO ka mea mua, ʻo nā kūlana ʻekolu-dimensional āpau o kēlā me kēia atom i loko o kēlā me kēia amino acid e koi nui i ka ʻimi. ʻO ka lua, hoʻohana nui nā protein i ka hoʻohui ʻana i kā lākou hana kemika e hoʻonohonoho pono i nā ʻātoma. No ka mea, loaʻa i nā polokina he mau haneli he mau haneli hydrogen paʻa "mea hāʻawi" (ʻo ka oxygen maʻamau) pono e pili kokoke i ka hydrogen bond "acceptor" (maʻa mau ka nitrogen i hoʻopaʻa ʻia i ka hydrogen), hiki ke paʻakikī loa ka loaʻa ʻana o nā conformations kahi kokoke i kēlā me kēia mea hāʻawi kokoke i ka mea ʻae. ʻO ke kolu, aia nā hiʻohiʻona liʻiliʻi no ka hoʻomaʻamaʻa ʻana i nā ʻano hoʻokolohua, no laila pono e hoʻomaopopo i ka hiki ke hoʻopili ʻia ʻekolu-dimensional ma waena o nā waikawa amino ma ke kumu o nā kaʻina 1D e hoʻohana ana i ka ʻike e pili ana i ka evolution o nā protein pili.

Ua hoʻohana mua ʻia ka physics e hoʻohālike i ka pilina o nā ʻātoma i ka ʻimi ʻana i ka conformation maikaʻi loa, a ua kūkulu ʻia kahi ala e wānana i ke ʻano o nā protein. Hāʻawi ʻia ʻo Karplus, Levitt a me Warshel i ka makana no ka 2013 Nobel no ka Chemistry no kā lākou hana ma ka simulation computational o nā protein. Eia nō naʻe, ʻoi aku ka pipiʻi o nā ʻano hana physics a koi ʻia i ka hoʻoili ʻana, no laila ʻaʻole hiki ke wānana ʻia nā ʻano hana ʻekolu. ʻO kekahi ala ʻē aʻe e pili ana i ka ʻike, ʻo ia ka hoʻohana ʻana i nā ʻikepili o nā hale i ʻike ʻia a me nā kaʻina e hoʻomaʻamaʻa i nā kumu hoʻohālike ma o ka naʻauao a me ka aʻo mīkini (AI-ML). Hoʻohana ʻo Hassabis a me Jumper i nā mea o ka physics a me AI-ML, akā ʻo ka hana hou a me ka lele ʻana i ka hoʻokō ʻana o ka hoʻokokoke mai AI-ML. Ua hoʻohui like nā mea noiʻi ʻelua i nā ʻikepili lehulehu me nā kumuwaiwai helu ʻenehana no ka hana ʻana iā AlphaFold.

Pehea mākou e ʻike ai ua "hoʻoponopono" lākou i ka puzzle wanana kūkulu? I ka makahiki 1994, ua hoʻokumu ʻia ka hoʻokūkū Critical Assessment of Structure Prediction (CASP), e hui ana i kēlā me kēia ʻelua makahiki e nānā i ka holomua o ka wānana ʻana. E kaʻana ka poʻe noiʻi i ke kaʻina 1D o ka protein nona ka hoʻolālā i hoʻoholo hou ai, akā ʻaʻole i paʻi ʻia nā hopena. Wahi a ka mea wānana i ke ʻano ʻekolu-dimensional me ka hoʻohana ʻana i kēia kaʻina 1D, a na ka mea loiloi e hoʻoponopono kūʻokoʻa i ka maikaʻi o nā hopena i wānana ʻia ma ka hoʻohālikelike ʻana iā lākou i ke ʻano ʻekolu-dimensional i hāʻawi ʻia e ka mea hoʻokolohua (i hāʻawi ʻia i ka mea loiloi wale nō). Mālama ʻo CASP i nā loiloi makapō ʻoiaʻiʻo a hoʻopaʻa i nā lele hana i kēlā me kēia manawa e pili ana i ka hana hou. Ma ka 14th CASP Conference ma 2020, ua hōʻike nā hopena wanana a AlphaFold i ka lele ʻana o ka hana i hoʻolaha ʻia e nā mea hoʻonohonoho ua hoʻoholo ʻia ka pilikia wānana 3D: ua kokoke ka pololei o ka hapa nui o nā wānana i nā ana hoʻokolohua.

ʻO ke koʻikoʻi ākea ʻo ia ka hana a Hassabis a me Jumper e hōʻike maopopo ana pehea e hiki ai iā AI-ML ke hoʻololi i ka ʻepekema. Hōʻike kāna noiʻi hiki iā AI-ML ke kūkulu i nā kuhiakau ʻepekema paʻakikī mai nā kumu ʻikepili he nui, hiki i nā mīkini nānā (e like me nā mea i loko o ChatGPT) ke ʻike i nā hilinaʻi nui a me nā hoʻoponopono ʻana i nā kumu ʻikepili, a hiki iā AI-ML ke hoʻoholo ponoʻī i ka maikaʻi o kāna mau hopena. Ke hana maoli nei ʻo AI-ML i ka ʻepekema.