Elix, CBI, 招待講演, ElixにおけるAI創薬と最新動向, 2021-10-26

Transcript

1. $PQZSJHIU
   ˜
   &MJY
   *OD
   "MM
   SJHIUT
   SFTFSWFE $#*ֶձ
   ট଴ߨԋ
   &MJYʹ͓͚Δ"*૑ༀͱ࠷৽ಈ޲ גࣜձࣾ&MJY $&0
   ݁৓৳࠸ 2021/10/26 1

2. $PQZSJHIU
   ˜
   &MJY
   *OD
   "MM
   SJHIUT
   SFTFSWFE ૑ༀʹ͓͚Δ՝୊ 2 Scannell et al. (2012) • 10೥Ҏ্ͷظؒɺ1000ԯԁҎ্

   ͷίετɺ੒ޭ཰ͷ௿͞ 
 • ૑ༀίετ͸ࢦ਺ؔ਺తʹ૿େ 
 • 2010೥Ҏ߱͸͜ͷ૿Ճ͕ࢭ·ͬͨ ͱ͢Δݚڀ΋͋Δ΋ͷͷґવͱ͠ ͯߴ͍ Eroom's Law

3. $PQZSJHIU
   ˜
   &MJY
   *OD
   "MM
   SJHIUT
   SFTFSWFE 3 3FUIJOLJOH
   %SVH
   %JTDPWFSZ
   ૑ༀΛ࠶ߟ͢Δ

4. "*ؔ࿈ຽؒ౤ࢿֹ
   ೥
   WT
   ೥ 4 https://aiindex.stanford.edu/report/ 2020೥Ͱ࠷΋େ͖͔ͬͨͷ͸૑ༀɻ͍ͭʹࣗಈӡసΑΓ΋େ͖ͳ౤ࢿֹʹɻ

5. $PQZSJHIU
   ˜
   &MJY
   *OD
   "MM
   SJHIUT
   SFTFSWFE "*ελʔτΞοϓͱ੡ༀձࣾͷఏܞ਺ 5 https://www.biopharmatrend.com/m/free-reports/ai/ ߹ܭఏܞ਺͸ॱௐʹ૿Ճ

6. "*ར༻༻్ͷ಺༁ 6 Schuhmancher et al. (2020) ௿෼ࢠ͕Ұ൪ଟ͍ ※ ੈքͷ੡ༀձࣾτοϓ21͕ࣾ2014ʙ2019 ʹ͔͚ͯऔΓ૊ΜͰ͍ͨAIϓϩδΣΫτ

7. 3FTUSJDUFE
   ˜
   &MJY
   *OD ෼ࢠઃܭ 7 Image Source: Sanchez-Lengeling et al. (2018)

   Drug-likeͳ෼ࢠ͸ʙ1060ݸ ࣮ݧ/γϛϡϨʔγϣϯ ༧ଌϞσϧ ੜ੒Ϟσϧ

8. $PQZSJHIU
   ˜
   &MJY
   *OD
   "MM
   SJHIUT
   SFTFSWFE &MJY
   %JTDPWFSZ5.
   "*
   %SVH
   %JTDPWFSZ
   1MBUGPSN 8 Elix͕ഓ͖ٕͬͯͨज़Λ݁ूͨ͠AI૑ༀϓϥοτϑΥʔϜ ᶃڞಉݚڀɹᶄϥΠηϯεఏڙ w ׆ੑ͋Γ
   w

   ׆ੑͳ͠
   w 
   O.
   w ʜ

9. $PQZSJHIU
   ˜
   &MJY
   *OD
   "MM
   SJHIUT
   SFTFSWFE &MJY
   %JTDPWFSZ5.
   "*
   %SVH
   %JTDPWFSZ
   1MBUGPSN 9 ࣗࣾݚڀΛؚΉ࠷৽ͷݚڀ੒ՌΛϓϥοτϑΥʔϜʹ࣮૷ɻࣗࣾݚڀͷྫΛަ͑ͭͭ঺հɻ ͜ͷޙͷεϙϯαʔηογϣϯͰ΋ϓϥοτϑΥʔϜΛ঺հ → ← CBIͰ΋5ͭޱ಄ൃද

   w ׆ੑ͋Γ
   w ׆ੑͳ͠
   w 
   O.
   w ʜ

10. $PQZSJHIU
    ˜
    &MJY
    *OD
    "MM
    SJHIUT
    SFTFSWFE &MJY
    1SFEJDUʢ༧ଌʣ 10 Properties Model • ׆ੑɺ෺ੑɺADMETͳͲΛ༧ଌ͢ΔϞδϡʔϧ • ैདྷϞσϧ͔Β࠷৽ͷάϥϑܥͷϞσϧ·Ͱ

    • ࣗಈͰϋΠύʔύϥϝʔλௐ੔Λͯ͠ϕετͳϞσϧΛબ୒ Case Studyɿ 
 Ξϯυϩήϯड༰ମʹର͢ΔόʔνϟϧεΫϦʔχϯά • ༧ଌ஋͚ͩͰͳ͘ɺCon fi denceείΞΛߟྀ͢Δ͜ͱͰ ΑΓ·ͱ΋ͳԽ߹෺Λબ୒ • ߜΓࠐΜͩ53ݸͷ͏ͪ34ݸ͕ಛڐऔಘࡁΈͷ΋ͷͩͬͨ 
 ʢ༧ଌϞσϧͷੑೳΛࣔ͢͜ͱ͕໨తͷ࣮ݧʣ Con fi denceείΞ͋Γ Con fi denceείΞͳ͠ 10/27 ޱ಄ൃද O4-3 Romeo Cozac 
 “Graph Convolutional Networks for Ligand-based Virtual Screening against the Androgen Receptor”

11. $PQZSJHIU
    ˜
    &MJY
    *OD
    "MM
    SJHIUT
    SFTFSWFE &MJY
    $SFBUFʢ෼ࢠੜ੒ʣ 11 Model • ੜ੒ϞσϧʹΑΓߏ଄ൃੜΛߦ͏Ϟδϡʔϧ • Elix Predictͱ૊Έ߹Θͤͯॴ๬ͷ׆ੑɾ෺ੑΛ࣋ͭ෼ࢠΛੜ੒

    • άϥϑϕʔεɺSMILESϕʔεɺϑϥάϝϯτϕʔεͳͲಠࣗϞσϧΛ ؚΉ༷ʑͳϞσϧΛαϙʔτ • ࢦఆͨ͠෦෼ߏ଄Λར༻͢Δ͜ͱ΋Մೳ eGEGLɿElixͷಠࣗϞσϧʢWüthrich et al. 2021ʣ • χϡʔϥϧωοτϫʔΫͱҨ఻తΞϧΰϦζϜΛ ߹ΘͤͨϞσϧ͕SOTAͱ͍͏എܠ • Ҩ఻తΞϧΰϦζϜ෦෼ʹυϝΠϯ஌ࣝΛೖΕΔ ͜ͱΛۃྗഉআͨ͠ΑΓҰൠԽͨ͠Ϟσϧ • ϕϯνϚʔΫͰಉ౳Ҏ্ Wüthrich et al. (2021) 10/27 ޱ಄ൃද O4-3 Pierre Wüthrich 
 “Using Attribution-based Explainability to Guide Deep Molecular Optimization”

12. $PQZSJHIU
    ˜
    &MJY
    *OD
    "MM
    SJHIUT
    SFTFSWFE 3FUSPTZOUIFTJTʢٯ߹੒ղੳʣ 12 Model • ٯ߹੒ղੳΛߦ͏Ϟδϡʔϧ • େྔͷԽ߹෺Λ·ͱΊͯॲཧ΋Մೳ •

    Elix CreateͰੜ੒͞ΕͨԽ߹෺ͷ߹੒༰қੑͷݕ౼ʹ΋ • ڞ௨͢ΔதؒମΛܦ༝͢Δϧʔτͷఏࣔ • Ձ֨ɺར༻Մೳͳࢼༀɺऩ཰౳Λߟྀͨ͠ϧʔτͷఏࣔ • ౦ژ޻ۀେֶͱͷڞಉݚڀ • ٯ߹੒ղੳπʔϧ܈ • Φʔϓϯιʔεͱͯ͠ެ։༧ఆ Elix Synthesize 10/27 ޱ಄ൃද O4-2 Haris Hasic 
 
 “RetroSynthWAVE: An Open- Source Software Platform for E ffi cient Chemical Synthesis Research”

13. $PQZSJHIU
    ˜
    &MJY
    *OD
    "MM
    SJHIUT
    SFTFSWFE σʔλ͕গͳ͍໰୊΁ͷରԠ 13 ࿈߹ֶशʢkMoL, Elix Milaʣ • ࿈߹ֶशɿσʔλΛ֎෦ʹग़͢͜ͱͳ͘ɺෳ਺اۀͷσʔλΛ׆ ༻ֶͯ͠श͢Δ͜ͱΛՄೳʹ͢Δٕज़

    • kMoL: ࿈߹ֶशͱ༧ଌϞσϧͰߏ੒͞ΕΔϥΠϒϥϦ • ژ౎େֶͱڞಉ։ൃ • ࿈߹ֶशϞδϡʔϧElix MilaΛϕʔεʹ։ൃ • Φʔϓϯιʔεͱͯ͠ϦϦʔεࡁΈ (https://github.com/elix-tech/kmol) ࣗݾڭࢣ͋ΓֶशʢSelf-Supervised Learningʣ • ࣮ݧ஋͕ଘࡏ͠ͳ͍Խ߹෺σʔλ΋׆༻ • ༧ଌਫ਼౓Λ޲্ͤ͞ΔΞϓϩʔν • σʔλ͕গͳ͍৔໘Ͱಛʹ༗༻ 10/27 ޱ಄ൃද O4-6 Laurent Dillard 
 “Improving Molecular Property Prediction using Self-supervised Learning”

14. $PQZSJHIU
    ˜
    &MJY
    *OD
    "MM
    SJHIUT
    SFTFSWFE ௿෼ࢠҎ֎ͷϞμϦςΟ 14 ௿෼ࢠ͕ϝΠϯͰ͋Δ΋ͷͷɺͦΕҎ֎ͷϞμϦςΟ΋ѻ͍ͬͯΔ ߅ମ H3ϧʔϓͷ3࣍ݩߏ଄༧ଌ • λϯύΫ࣭഑ྻσʔλʹΑΔڭࢣͳֶ͠शΛ׆༻ •

    ͜ͷಛ௃நग़ʹΑΓH3ϧʔϓߏ଄༧ଌͷਫ਼౓Λ޲্ 10/27 ޱ಄ൃද O4-4 David Jimenez 
 “Leveraging Self-Supervised Contextual Language Models for DNN Antibody CDR-H3 Loop Predictions”

15. $PQZSJHIU
    ˜
    &MJY
    *OD
    "MM
    SJHIUT
    SFTFSWFE ϝϯόʔ 15 🌏 ੈքத͔Β༏लͳϝϯόʔΛ࠾༻ • AIݚڀऀɺAIΤϯδχΞ • έϛετɺόΠΦϩδετ

    • ത࢜߸औಘऀଟ਺ • ӳޠެ༻ޠ 🇯🇵 🇭🇺 🇫🇷 🇦🇺 🇲🇾 🇰🇬 🇺🇸 🇳🇬 🇨🇭 🇧🇦 🇨🇴 🏢 ૑ۀऀ ݁৓৳࠸ɺPh.D 
 ڞಉ૑ۀऀɾCEO େٱอୡ໼ ڞಉ૑ۀऀɾCOO ౦ژʹू·ͬͯ࢓ࣄ Λ͍ͯ͠·͢ 📍

16. Thank you! Contact us: info@elix-inc.com