search

bulik / ldsc

LD Score Regression (LDSC)
GNU General Public License v3.0
652 stars 345 forks source link

[Cell-type specific analysis] How are the 'control' gene sets determined? #136

Open mariacsplc opened 6 years ago

mariacsplc commented 6 years ago

Hi, I am trying to run the partition heritability functionality with my own annotations. I managed to reproduce your results from Finucane 2015, on the Cahoy annotations. I was wondering if you could help clarify how the control genes are determined for generating the control ldscore files for the annotations- are they all consisting of all the expressed genes in the analysed data set? Or is there I am taking about files referred to in here: https://github.com/bulik/ldsc/wiki/Cell-type-specific-analyses under --ref-ld-chr-cts. Thanks a lot!

hilaryfinucane commented 5 years ago

Hi, The control gene sets are all genes that were included in the analysis to determine differential expression. For the Cahoy annotations, these were downloaded from supplemental tables from the Cahoy et al paper.

Best, Hilary

On Thu, Nov 15, 2018 at 11:35 AM mariacsplc notifications@github.com wrote:

Hi, I am trying to run the partition heritability functionality with my own annotations. I managed to reproduce your results from Finucane 2015, on the Cahoy annotations. I was wondering if you could help clarify how the control genes are determined for generating the control ldscore files for the annotations- are they all consisting of all the expressed genes in the analysed data set? Or is there I am taking about files referred to in here: https://github.com/bulik/ldsc/wiki/Cell-type-specific-analyses under --ref-ld-chr-cts. Thanks a lot!

— You are receiving this because you are subscribed to this thread. Reply to this email directly, view it on GitHub https://github.com/bulik/ldsc/issues/136, or mute the thread https://github.com/notifications/unsubscribe-auth/AHSzHGqkqoTNCVwv8Ajj--WNDQpCrkj8ks5uvZfTgaJpZM4YgXTz .

biostat0903 commented 5 years ago

Hi, I hope to check that question. If I use the LDSC to analysis RNA-seq data of different tissues, without any control information. When I only input the annotation of one tissue to LDSC, it can also output a result. Would you like to tell me whether the process is right. Is the "control" data specific to Cahoy data? Many thanks for your kind help! Best, Sheng

hilaryfinucane commented 5 years ago

Hi Sheng,

Sorry for the lack of clarity, I guess "control" can mean a lot of different things. In this case what I mean is: any time you use LDSC to assess a particular gene set for enrichment, you want to also include the larger gene set those genes were chosen from, so that you are not just getting inflated scores from the fact that variants near genes in general tend to have more activity. So if you have RNA-seq from several tissues, then you want to include the set of all genes that passed QC as your "control" gene set, and then the set of specifically expressed genes in the tissue of interest as your main annotation.

Best, Hilary

On Wed, Dec 26, 2018 at 9:09 PM Sheng Yang notifications@github.com wrote:

Hi, I hope to check that question. If I use the LDSC to analysis RNA-seq data of different tissues, without any control information. When I only input the annotation of one tissue to LDSC, it can also output a result. Would you like to tell me whether the process is right. Is the "control" data specific to Cahoy data? Many thanks for your kind help! Best, Sheng

— You are receiving this because you commented. Reply to this email directly, view it on GitHub https://github.com/bulik/ldsc/issues/136#issuecomment-450055890, or mute the thread https://github.com/notifications/unsubscribe-auth/AHSzHHusQUErQtUSSUQKZ264DDjZ7JTdks5u9CvcgaJpZM4YgXTz .