Riba:
- Gyaran gwaje-gwaje da yawa (gyara p-darajar) a cikin nazarin magana daban-daban da ikon fassara canjin ninka daidai da guje wa abubuwan karya.
- Gano tasirin tsari da ƙara shi zuwa ƙirar tare da PCA da raba hayaniyar fasaha da bambancin ilimin halitta
- Ikon danganta kowane asalin tafarki zuwa tushe da sarrafa shi tare da daidaiton ilimin halitta a cikin haɓakar hanya da haɗakarwa da yawa-omics
Tantanin halitta ba lamba ɗaya ba ce; Tsari ne inda dubban kwayoyin halitta, sunadaran sunadarai da metabolites suke rawa lokaci guda. Omics (sunan gama-gari don hanyoyin da ke auna Layer Layer gaba ɗaya) yayi ƙoƙarin kama wannan rawa duka: genomics (DNA), transcriptomics (RNA - wanda kwayoyin halitta suke aiki da nawa), proteomics (proteins), metabolomics (kananan kwayoyin halitta). Kowane Layer omics yana samar da dubban bayanai masu girma, hayaniya da tsada. AI yana da ƙarfi a bincikar wannan babban girman bayanai da alamu; Amma kai ne wanda ke yanke shawarar wane tsari shine gaskiyar halitta kuma wane hayaniyar fasaha ce.
A cikin wannan rukunin, za mu ci gaba ta hanyar kwafi, mafi yawan binciken omics; Ka'idodin sun shafi sauran yadudduka kuma. Yawan aiki na yau da kullum: matrix na magana daga bayanan da aka samo ( layuka sune kwayoyin halitta, ginshiƙai sune samfurori, sel sune matakan magana), daidaitawa (cire bambance-bambancen fasaha), nazarin maganganu daban-daban (neman kwayoyin halittar da ke canzawa tsakanin yanayi guda biyu), wadatar hanyar hanya (gano wace hanyoyin nazarin halittu da canje-canjen kwayoyin halitta suka taru) da fassarar.
Bambance-bambancen magana: ninka canji da gyara p-darajar
Don sanin idan kwayar halitta ta "canza," ana duba lambobi biyu: sauyin ninka - sau nawa magana yana ƙaruwa / raguwa, yawanci akan ma'auni na log2 - da kuma daidaita p-darajar (padj - kididdigar da ke sarrafawa don ƙimar ƙarya lokacin da aka yi gwaji da yawa). Me yasa gyara? Domin kuna gwada kwayoyin halitta 20,000 a lokaci guda; Ko da kwatsam, ɗaruruwan kwayoyin halitta na iya zama “mahimmanci”. Ba tare da gyare-gyaren gwaji da yawa ba - iyakance ƙimar gano ƙarya tare da hanyoyi irin su Benjamini-Hochberg - jerin suna yaudara. AI na iya rubuta rubutun da ke ƙididdige wannan ƙididdiga, amma idan ya tsallake gyara, sakamakon ku ba shi da kariya a kimiyyance.
Tsanaki: Wani AI na iya cewa "kwayoyin halitta 500 sun canza sosai" bisa tushen p-darajar. Duban p-darajar da aka gyara, lambar zata iya raguwa zuwa 30. Koyaushe bincika gyaran gyare-gyaren gwaji da yawa; Wannan shine bambanci tsakanin karɓa da ƙin buga littafin.
Tasirin tsari: tarko mafi ban tsoro
Tasirin tsari (bambancin fasaha da ke tasowa daga sarrafa samfuran ta kwanaki daban-daban, na'urori ko mutane) shine babban tushen kuskure a cikin binciken omics. Idan an sarrafa yanayin ku guda biyu a cikin kwanaki biyu daban-daban, "bambancin nazarin halittu" da kuke gani na iya zama ainihin bambancin rana. AI na iya ba da shawarar ƙara madaidaicin tsari zuwa ƙirar (misali ~ yanayin yanayin), amma alhakinku ne don saita shi daidai kuma kada ku haɗa shi cikin ƙirar gwaji.
Tukwici: Kafin fara bincike, zana PCA (Hanyar Nazari na Farko - hanyar da ke taƙaitawa da hango manyan bayanai akan gatari da yawa). Idan samfurori sun taru ta tsari maimakon ta yanayin halitta, tasirin batch yana da rinjaye kuma dole ne a gyara shi da farko.
Haɗin kai Multi-omics
Fahimtar gaske sau da yawa tana zuwa ne daga haɗa sassan layi ɗaya: idan kwayar halitta tana aiki tuƙuru amma sunadarin sa baya ƙaruwa, ƙa'ida tana a matakin fassarar. Haɗin kai Multi-omics-haɗa nau'ikan omics daban-daban a cikin samfuri ɗaya - shine inda AI ke samun ƙarfi amma kuma inda ya fi ɓatar da; saboda ma'auni, hayaniya da samfurin ma'auni na yadudduka sun bambanta. AI yana ba da shawarar aikin haɗin kai, amma kuna sarrafa daidaiton nazarin halittu na sakamakon.
uku mini lokuta
Case 1 - Haɓaka haɓakawa. 1,240 nau'in jinsin halittu daban-daban an samo su a cikin aikin kansa. AI ya ƙaddamar da su don wadatar da hanya, nuna alamar zagayowar tantanin halitta da hanyoyin gyaran DNA; Ƙungiyar ta ƙirƙiri taswirar hasashe a cikin sa'o'i 2. Amma sun sake gwada kowace hanya tare da kayan aiki mai zaman kansa (g:Profiler) kuma sun gano cewa hanya ɗaya ta AI ta kuskure.
Case 2 - Batch tarko. Ɗaya daga cikin dakin gwaje-gwaje ya sami "bambanci" 900-bambanci tsakanin kungiyoyin jiyya guda biyu. Lokacin da suka yi PCA, sun ga cewa an raba samfuran ta hanyar jeri tsari. Bayan gyaran tsari, ainihin bambancin ya ragu zuwa kwayoyin halitta 60. AI ba da niyya ba ta tsallake canjin tsari a farkon bincike.
Case 3 - Sunan hanyar da aka yi. Wani dalibi ya ba AI jerin kwayoyin halitta kuma ya tambaya, "Wace hanya ce ta KEGG?" AI ta ba da ID na hanya da suna kamar gaske. Lokacin da ɗalibin ya bincika KEGG, ya ga cewa ID ɗin ba ya wanzu; tabbatarwa ya hana sakamakon ƙirƙira.
Samfura huɗu masu kwafi
1) DESeq2 tsarin tafiyar aiki:
Matsayinku: Masanin ilimin lissafi. Rubuta rubutun mataki-mataki don nazarin maganganun bambancin RNA-seq tare da R/DESeq2: ƙidaya karatun matrix, ƙirar ƙira (~ tsari + yanayin), daidaitawa, tebur sakamako. Yi amfani da gyare-gyaren gwaji da yawa (BH) da amfani da padjcolumn. Bayyana abin da kowane mataki yake yi a cikin layin sharhi.
2) Kyakkyawan / sarrafa tsari:
Ka ba ni jerin abubuwan bincike na RNA-seq QC: sarrafa tsari tare da PCA, girman ɗakin karatu, adadin gano kwayoyin halitta, ganowa a waje. Ga kowane ma'auni, saka madaidaicin "abin da na gani ya sa ni damuwa". Bayyana abin da ya kamata in yi idan tsari da yanayin halitta sun haɗu.
3) Tabbatar da sakamakon haɓakawa:
Zan ba ku jerin hanyoyin da aka wadatar (yawan hanyoyin, padj, genes). Rubuta ainihin kowane hanya (KEGG/GO ID) a zahiri kuma kar a daidaita shi. Tace sakamakon da padj <0.05. Ƙayyade waɗanne hanyoyi ne ke tallafawa juna ta hanyar ilimin halitta, amma sanya kowane ainihi a matsayin "dole ne a tabbatar da su a cikin bayanan bayanai."
4) Duban daidaiton Multi-omics:
Rubutun rubutu da furotin suna haifar da sabani na bayanin kwatancen nau'in halitta/na gina jiki. Yi lissafin yuwuwar nazarin halittu (gyaran fassarar bayan fassarorin) da fasaha (hayaniyar aunawa, daidaitaccen samfurin) dalilan wannan kuma gaya mani yadda zan gwada kowane.
Rauni mai ƙarfi / Ƙarfi mai ƙarfi
Rawanin faɗakarwa:
Sunan mahimman hanyoyi a cikin wannan jerin kwayoyin halitta.
Babu tushe, babu ƙididdiga, babban haɗarin ƙirƙira hanyoyin.
Ƙarfi mai ƙarfi:
Matsayinku: Masanin ilimin lissafi. A cikin tebur bambancin jinsin da aka haɗe (gene, log2FC, padj) kawai ɗauki kwayoyin halitta tare da padj <0.05. Faɗa mani matakan binciken haɓakar GO da za a yi tare da waɗannan kwayoyin halitta da kayan aiki (g: Profiler) zan yi amfani da su. Sunan hanyar shine FAKE; Zan gudanar da kayan aiki kuma in yi bincike, kawai ku bayyana madaidaicin hanya da gyaran gwaji da yawa.
Bambanci: bayyanannen tacewa, mayar da hankali kan hanya, hana ƙirƙira da barin tabbaci ga mai amfani.
Matakan nazarin Omics
mataki
Manufar
kuskure na kowa
AI rawar
al'ada
Kawar da bambancin fasaha
Zaɓin hanyar da ba daidai ba
Rubutun + dalili
PCA/QC
Gano tsari da waje
tsallake mataki na
Hoto + sharhi
kalamai daban-daban
Nemo canjin kwayoyin halitta
Ba a gyara p
Daftarin rubutun
wadata
nemo hanya
hanyar ƙirƙira
hanya
hadewa
hade yadudduka
Kuskuren sikeli/match
Shawarar kwararar aiki
Single cell omics: sabon sikelin
A cikin 'yan shekarun nan, jerin tantanin halitta guda - auna bayanin bayanin kowane dubunnan tantanin halitta daban - ya ɗauki omics zuwa sabon girma. Yanzu, maimakon "matsakaicin magana na nama", za mu iya ganin kowane nau'in tantanin halitta a cikin wannan nama daban. Wannan ikon yana gabatar da sabbin ramuka: bayanan ba su da yawa (mafi yawan kwayoyin halitta ba su da karatun sifili a cikin mafi yawan sel-dropout), girman dubun duban sel × dubu ashirin da dubunnan kwayoyin halitta, kuma rarrabuwar nau'ikan tantanin halitta galibi ana yin su ta hanyar tari. AI yana da ƙarfi wajen samar da tari da nau'in alamar tantanin halitta akan bayanan tantanin halitta guda ɗaya; amma kuna tabbatar da sanannun kwayoyin halitta ko kowane gungu ainihin nau'in tantanin halitta ne ko kayan fasaha (misali matattun ƙwayoyin cuta, sel biyu da aka kama tare). Kar a karɓi lakabin nau'in tantanin halitta wanda AI ya ba da shawara ba tare da tabbatar da alamun alamun wannan tarin a cikin ainihin wallafe-wallafe ba.
Tukwici: A cikin nazarin tantanin halitta guda ɗaya, idan AI ta ba da shawarar alamar “T cell” zuwa gungu, bincika wa kanku cewa alamun T cell (misali CD3) an bayyana su sosai a cikin wannan gungu. Idan alamar ba ta goyan bayan alamar ba, hasashe ne, ba ƙarshe ba.
Kuskuren gama gari
- Tsallake gyaran gwaji da yawa. Lissafin ya kumbura tare da p-darajar raw; ya kamata a yi amfani da padj.
- Kuskure tasirin tsari ga ilmin halitta. Ya kamata a fara duba shi tare da PCA.
- Yin bene canza ma'auni kawai. Canjin babban ninki na iya zama ɓata a cikin ƙananan maganganu, ƙwayoyin hayaniya.
- Karɓar hanyar da aka yi gyare-gyare/ ainihi. Dole ne a tabbatar da kowane ainihi a cikin ma'ajin bayanai.
- Rashin kimanta girman samfurin. Ƙarfin ƙididdiga yana da ƙasa a cikin ƙirar 2 ta 2; Ya kamata a fassara sakamakon da hankali.
a takaice
Binciken Omics yana aiki tare da babban girma, bayanai masu hayaniya, kuma AI yana haɓaka wannan bayanan ta hanyar bincika shi, rubuta rubutun, da alamu masu alama. Amma gyare-gyaren gwaji da yawa a cikin furci na daban, sarrafa tsari tare da PCA da tabbatar da tushe a cikin wadata suna da mahimmanci. Haɗin kai Multi-omics yana da ƙarfi amma yaudara; Bincika kowane sakamako tare da daidaiton ilimin halitta, la'akari da ma'auni da bambance-bambancen amo na yadudduka.
Aikin aikace-aikace
Nemo matrix ƙidayar RNA-seq na jama'a (misali daga GEO). Shin AI ta rubuta rubutun bincike tare da samfurin "DESeq2 workflow" kuma bincika lamba cewa an yi amfani da gyaran gwaji da yawa. Sannan nemi AI don sharhin haɓakawa kuma tabbatar da duk ID ɗin hanyar da yake dawowa ɗaya bayan ɗaya akan bayanan KEGG ko GO; Ka lura da nawa ne na gaske.
jerin abubuwan dubawa
- [] Na yi amfani da padj (gyara) a cikin bambance-bambancen bincike, ba raw p-darajar ba.
- [ ] Na duba tasirin batch tare da PCA kuma na ƙara shi zuwa samfurin idan ya cancanta.
- [ ] Na fassara kwayoyin halitta tare da babban canjin ninka amma ƙananan magana tare da taka tsantsan.
- [ ] Na tabbatar da kowace hanya/ID ID a kan ainihin bayanan bayanai.
- [] Na kimanta ikon ƙididdiga na girman samfurin.
- [ ] Na raba sabani da yawa-omics zuwa dalilai na halitta da fasaha.