Implement cell type deconvolution of bulk datasets

[alexvpickering]

This is something I have been doing with Alexi/Grant/Rashmi that I feel is very useful for bulk datasets. It uses marker genes from a comparable single cell dataset to estimate the proportion of different cell types within bulk samples. I have been using dtangle.

It makes sense to put this in the Datasets tab so that more than two groups can be selected (e.g. control, low ferritin, high ferritin) as opposed to requiring only a test and control group.

[ikohane]

Great. How can we see this?

On Nov 6, 2019, at 9:43 AM, Alex Pickering notifications@github.com wrote:

This is something I have been doing with Alexi/Grant/Rashmi that I feel is very useful for bulk datasets. It uses marker genes from a comparable single cell dataset to estimate the proportion of different cell types within bulk samples. I have been using dtangle https://urldefense.proofpoint.com/v2/url?u=https-3A__github.com_gjhunt_dtangle&d=DwMCaQ&c=WO-RGvefibhHBZq3fL85hQ&r=LGvMyVydq3L28lQxe97sG_94kjwVf2ra9cq7q2wvXa0&m=nxTiAiofvfHowuUiTnRr4XstTS3n-jgTzEMcKNU6onA&s=B76MZQvynbxneNEz2abz45uAECMczIqplhxViC3R6NE&e=.

It makes sense to put this in the Datasets tab so that more than two groups can be selected (e.g. control, low ferritin, high ferritin) as opposed to requiring only a test and control group.

— You are receiving this because you are subscribed to this thread. Reply to this email directly, view it on GitHub https://urldefense.proofpoint.com/v2/url?u=https-3A__github.com_hms-2Ddbmi_drugseqr_issues_139-3Femail-5Fsource-3Dnotifications-26email-5Ftoken-3DACICXBBZ5KS5WYKFCKRCOITQSLJ2RA5CNFSM4JJV5GCKYY3PNVWWK3TUL52HS4DFUVEXG43VMWVGG33NNVSW45C7NFSM4HXH6GMQ&d=DwMCaQ&c=WO-RGvefibhHBZq3fL85hQ&r=LGvMyVydq3L28lQxe97sG_94kjwVf2ra9cq7q2wvXa0&m=nxTiAiofvfHowuUiTnRr4XstTS3n-jgTzEMcKNU6onA&s=ILns9EHSELEPFq6kWZiwbtKzF29n9QdlhaFfW39cMFU&e=, or unsubscribe https://urldefense.proofpoint.com/v2/url?u=https-3A__github.com_notifications_unsubscribe-2Dauth_ACICXBFSRXZQETQGM7NOT7LQSLJ2RANCNFSM4JJV5GCA&d=DwMCaQ&c=WO-RGvefibhHBZq3fL85hQ&r=LGvMyVydq3L28lQxe97sG_94kjwVf2ra9cq7q2wvXa0&m=nxTiAiofvfHowuUiTnRr4XstTS3n-jgTzEMcKNU6onA&s=JOE4hBPum_r1lAZN_BKYkRUigRCrIFXKyx0-fMT4QD4&e=.

[alexvpickering]

Not yet implemented but this is an example of a one off analysis I did for Alexi/Grant:

Each point is a bulk RNA-Seq sample showing the estimated proportion of the cell type on the x-axis. Here is the interpretation:

Cellular deconvolution of the isolated CD14+ monocytes using marker genes for classical and alternatively-activated monocytes (from a single cell PBMC dataset) suggests a generally increasing proportion of classically activated, pro-inflammatory M1 macrophages as compared to alternatively activated anti-inflammatory M2 macrophages going from control to low ferritin to high ferritin samples.

[alexvpickering]

closed with 82f3647bb4dd23bbad4f78a5b79b1a0019311258. This is what it looks like:

Will let you know once live @ikohane.

[alexvpickering]

Forgot to mention it @ikohane but it's live on either drugseqr.com or sjia.drugseqr.com