search

MaayanLab / sigcom-lincs

Signature Commons LINCS Repo
3 stars 3 forks source link

Missing LINCS L1000 Chemical Perturbagens library on result page #80

Closed heesukim628 closed 9 months ago

heesukim628 commented 9 months ago

On the search results page, I can't find the LINCS L1000 Chemical Perturbagens library. @AviMaayan @jeevangelista

jeevangelista commented 9 months ago

can you give more info/screenshots?

heesukim628 commented 9 months ago

From Playbook use case 3 there are stage 'Extract LINCS L1000 Chemical Perturbagens - Resolved drugs from the LINCS L1000 Chemical Perturbagens library'. And from the result page of SigCom LINCS, I can't find the perturbagens library card. I checked other perturbations card but couldn't find what has the same result as playbook shows.

Below are inputs and outputs from playbook.

[Up Genes] ABCB5 CA5A CACNA2D3-AS1 CDKN2A CHL1-AS1 EDA2R EYA4 FGF5 FGF6 HNRNPMP1 KCNG4 LINC01479 LMO3 LOC101927023 MTHFD2P1 PTCHD4 PYHIN1 SCGB3A2 SPATA18 TUFMP1 ZMAT3

[Down Genes] AGTR2 CLP1 FASTKD5 GDF3 GML HPS6 LINC00911 LOC102724710 MAD2L1BP OTUD6A PCAT4 RIT2 SFTPA1 SMG8 STRA8 TEX13C TUBB2BP1 ZNF518B

[Result] image

u8sand commented 9 months ago

It's this one? image

heesukim628 commented 9 months ago

That library shows a Z score of ZM-241385 as -5.043 but playbook table shows a Z score of 2.56? image

u8sand commented 9 months ago

Okay, it works properly with gene sets image

But it doesn't seem to work properly with up/down sets image

AviMaayan commented 9 months ago

[like] Maayan, Avi reacted to your message:

From: Daniel Clarke @.> Sent: Tuesday, November 28, 2023 4:38:29 PM To: MaayanLab/sigcom-lincs @.> Cc: Maayan, Avi @.>; Mention @.> Subject: Re: [MaayanLab/sigcom-lincs] Missing LINCS L1000 Chemical Perturbagens library on result page (Issue #80)

USE CAUTION: External Message.

Okay, it works properly with gene sets image.png (view on web)https://urldefense.proofpoint.com/v2/url?u=https-3A__github.com_MaayanLab_sigcom-2Dlincs_assets_1341861_70f76960-2D8cc1-2D4485-2Da574-2De56c6f9898ce&d=DwMCaQ&c=shNJtf5dKgNcPZ6Yh64b-ALLUrcfR-4CCQkZVKC8w3o&r=OTriZuxvm8goxicZHLrvnuKigY-wW0EZRYZ7nbS2vys&m=bnRagf_P09zP4wzF2MO3ACR0LzaRZy6HO_Y8EinQocHQnSIWjAP48gth3dABO36Y&s=xDburpL2KVgna-_ZAaN-wyTcFNIV2voEhRXKK8RlG4s&e=

But it doesn't seem to work properly with up/down sets image.png (view on web)https://urldefense.proofpoint.com/v2/url?u=https-3A__github.com_MaayanLab_sigcom-2Dlincs_assets_1341861_55ba33e2-2Dd98b-2D4273-2Da614-2D7f7402b24c88&d=DwMCaQ&c=shNJtf5dKgNcPZ6Yh64b-ALLUrcfR-4CCQkZVKC8w3o&r=OTriZuxvm8goxicZHLrvnuKigY-wW0EZRYZ7nbS2vys&m=bnRagf_P09zP4wzF2MO3ACR0LzaRZy6HO_Y8EinQocHQnSIWjAP48gth3dABO36Y&s=345In2dqcvVc4nItOo0sN8ypFVSYmoE8TJHr0aRhbyQ&e=

— Reply to this email directly, view it on GitHubhttps://urldefense.proofpoint.com/v2/url?u=https-3A__github.com_MaayanLab_sigcom-2Dlincs_issues_80-23issuecomment-2D1830264706&d=DwMCaQ&c=shNJtf5dKgNcPZ6Yh64b-ALLUrcfR-4CCQkZVKC8w3o&r=OTriZuxvm8goxicZHLrvnuKigY-wW0EZRYZ7nbS2vys&m=bnRagf_P09zP4wzF2MO3ACR0LzaRZy6HO_Y8EinQocHQnSIWjAP48gth3dABO36Y&s=OAn_mcQdp-jYeAu9GV6ZxCT-_JTA1hlFW3OzsqOnWiQ&e=, or unsubscribehttps://urldefense.proofpoint.com/v2/url?u=https-3A__github.com_notifications_unsubscribe-2Dauth_ADBIR2L3S7QNIZEJKZRRTY3YGYHQLAVCNFSM6AAAAAA755NQ4CVHI2DSMVQWIX3LMV43OSLTON2WKQ3PNVWWK3TUHMYTQMZQGI3DINZQGY&d=DwMCaQ&c=shNJtf5dKgNcPZ6Yh64b-ALLUrcfR-4CCQkZVKC8w3o&r=OTriZuxvm8goxicZHLrvnuKigY-wW0EZRYZ7nbS2vys&m=bnRagf_P09zP4wzF2MO3ACR0LzaRZy6HO_Y8EinQocHQnSIWjAP48gth3dABO36Y&s=idXtUXoN-P-l_VLca1lP3CEHjG-hEhl93DbyWCNYrRk&e=. You are receiving this because you were mentioned.Message ID: @.***>

u8sand commented 9 months ago

Confirmed this is an issue with the playbook and not sigcom-lincs

Basically it has to do with how we're turning two z scores (z-up/z-down) into one. will fix the calculation to be more intuitive

u8sand commented 9 months ago

These are now aligned by using sigcom-lincs data API's z-sum field. We'll want to discuss which method we want to use to compute z-sum