The way to calculate T cell dysfunction score

[Fanglv1990]

Hello,

I noticed that the description of result downloaded from your website mentioned that the z score is T cell dysfunction score. The calculation method of dysfunction score and exclusion score in your artical is the pearson correlation between the tumor expression profiles and TIDE signatures of T-cell dysfunction and exclusion, respectively. But in your script, the T cell dysfunction score is calculated by coxph. What kind of test method should be used ? What signatures should be used for calculate exclusion score?

Best wishes, Fang

[foreverdream2]

Hi, Fang,

Please notice that there are two types of scores: the first set of score is on the gene level. We use the scripts (based on CoxPH regression) for computation. The second set of score is a prediction value on the patient level. We used the pearson correlation for computation discussed in our paper. The signatures for computing patient-level exclusion score is available at our download page: http://tide.dfci.harvard.edu/download/Results.zip

Hope these points could help you.

Btw, I don't check this email often. Please use another email for future communication: pengj@alumni.princeton.edu

Sincerely, Peng

On Fri, Jul 5, 2019 at 7:29 AM Fanglv1990 notifications@github.com wrote:

Hello,

I noticed that the description of result downloaded from your website mentioned that the z score is T cell dysfunction score. The calculation method of dysfunction score and exclusion score in your artical is the pearson correlation between the tumor expression profiles and TIDE signatures of T-cell dysfunction and exclusion, respectively. But in your script, the T cell dysfunction score is calculated by coxph. What kind of test method should be used ? What signatures should be used for calculate exclusion score?

Best wishes, Fang

— You are receiving this because you are subscribed to this thread. Reply to this email directly, view it on GitHub https://github.com/foreverdream2/dysfunction_interaction_test/issues/1?email_source=notifications&email_token=ADTJLHE4AM66BK4XYRKJIQTP54WBHA5CNFSM4H6KPKB2YY3PNVWWK3TUL52HS4DFUVEXG43VMWVGG33NNVSW45C7NFSM4G5RGFBQ, or mute the thread https://github.com/notifications/unsubscribe-auth/ADTJLHDDOSTKC2CCXCMMPDDP54WBHANCNFSM4H6KPKBQ .

[Fanglv1990]

Hi Peng,

Thank you for your help. But I still can't find exclusion signatures. As I understand it, the TIDE signatures is a genome-wide vector. I just find a column named 'T Dysfunction' in a file named 'all_scores', but I can't find any information like these for exclusion signatures. I noticed that you provide a table of log-fold change of genes in each exclusion signatures. How can I calculate TIDE prediction scores for patients using these values?

Best wishes, Fang

[foreverdream2]

Hi, Fang,

After downloading the result package, you should see a folder named " Exclusion_scores", which contains the T-cell exclusion signatures. The mean column is the average signature utilized to calculate patient-level exclusion score.

On Mon, Jul 8, 2019 at 12:07 AM Fanglv1990 notifications@github.com wrote:

Hi Peng,

Thank you for your help. But I still can't find exclusion signatures. As I understand it, the TIDE signatures is a genome-wide vector. I just find a column named 'T Dysfunction' in a file named 'all_scores', but I can't find any information like these for exclusion signatures. Is it possible for me to calculate TIDE score by the information(signatures ) provided by you?

Best wishes, Fang

— You are receiving this because you commented. Reply to this email directly, view it on GitHub https://github.com/foreverdream2/dysfunction_interaction_test/issues/1?email_source=notifications&email_token=ADTJLHGFGA2OWJLWLU4GAL3P6K4Q3A5CNFSM4H6KPKB2YY3PNVWWK3TUL52HS4DFVREXG43VMVBW63LNMVXHJKTDN5WW2ZLOORPWSZGODZL5AWA#issuecomment-509071448, or mute the thread https://github.com/notifications/unsubscribe-auth/ADTJLHDTK3MI2H43D73PLYDP6K4Q3ANCNFSM4H6KPKBQ .

[Fanglv1990]

Hi， Peng，

Sorry, the file I downloaded is incomplete. After I redownloaded the result package, I find the folder you metioned. I find that there is one exclusion scores for one gene , but it looks like the dysfunction score is cancer-specific. Should I calculate TIDE score by the value of 'T Dysfunction' of specific cancer type? You selected 770 TIDE.selected genes. Should I calculate TIDE sore by the TIDE signatures of these genes? Where can I find the list of these genes?

Best wished, Fang

[benstaf]

Hello @foreverdream2 , I can't calculate the T cell dysfunction score for my dataset (the Web app gives the TIDE score, but I only want the dysfunction score), I am a bit confused, like @Fanglv1990.

I also found the column named 'T Dysfunction' in a file named 'all_scores' in http://tide.dfci.harvard.edu/download/Results.zip, but this is a column of length: 2997789, while there are only ~20 000 genes in humans.

So how do I compute the T dysfunction score for my dataset from there?

[data2intelligence]

Hi Peng,

Thank you for your help. But I still can't find exclusion signatures. As I understand it, the TIDE signatures is a genome-wide vector. I just find a column named 'T Dysfunction' in a file named 'all_scores', but I can't find any information like these for exclusion signatures. I noticed that you provide a table of log-fold change of genes in each exclusion signatures. How can I calculate TIDE prediction scores for patients using these values?

Best wishes, Fang

Hi, Fang, sorry for not replying to you for a long time, as after I left, I never used this Github account. For your question, yes, the T dysfunction scores are cancer specific but exclusion scores are static and do not change across cancer types. If possible, please use all genes, not only these 770 from NanoString panels, to make predictions. We selected 770 genes to fit the NanoString data analysis. But if you have a genome-wide transcriptomic profile, please don't restrain within this subset.

[data2intelligence]

Hello @foreverdream2 , I can't calculate the T cell dysfunction score for my dataset (the Web app gives the TIDE score, but I only want the dysfunction score), I am a bit confused, like @Fanglv1990.

I also found the column named 'T Dysfunction' in a file named 'all_scores' in http://tide.dfci.harvard.edu/download/Results.zip, but this is a column of length: 2997789, while there are only ~20 000 genes in humans.

So how do I compute the T dysfunction score for my dataset from there?

@benstaf Hi, Ben, sorry for my late response, as after I left my postdoc institute, I never used this Github account. Please see my response to Fang above. Also, the T-cell dysfunction gene scores are different from the TIDE ICB response prediction. To compute T dysfunction scores for your own dataset, please use the source code in this Github package.