Vignette 1 with CAF-S1 and S4 numbers aren't matching

[zkatmich]

Hello, I was running through the first vignette with the cytokine interaction in CAF-S1 and S4 and noticed my numbers in the figures aren't the same as shown in the vignette figures. It seems mostly to be a scaling issue but the p-value heat map was significantly different. Any feedback or assistance would be greatly appreciated. Thanks in advance.

[lmassenet-regad]

Hello, I would be happy to help but I need a bit of context/explanation of what you did and current issues to identify what is happening...

[zkatmich]

Yes, I downloaded the required files for the database, CC, and PC information from the Github and proceeded to copy the R code in the CAF vignette and run as I worked my way through it. I noticed in particular that the majority of the plots did not match the ones depicted in the vignette. I have included the versions I generated for reference. The main variation were the values in the balloon plot, several of the bar plots having widely different distributions, and the heatmap having quite different regions. Let me know if you need more information. Thanks for your help.

---

From: Lucile Massenet-Regad @.> Sent: Friday, March 31, 2023 10:18 AM To: soumelis-lab/ICELLNET @.> Cc: zkatmich @.>; Author @.> Subject: Re: [soumelis-lab/ICELLNET] Vignette 1 with CAF-S1 and S4 numbers aren't matching (Issue #11)

External Email - Use Caution

Hello, I would be happy to help but I need a bit of context/explanation of what you did and current issues to identify what is happening...

— Reply to this email directly, view it on GitHubhttps://nam02.safelinks.protection.outlook.com/?url=https%3A%2F%2Fgithub.com%2Fsoumelis-lab%2FICELLNET%2Fissues%2F11%23issuecomment-1491996569&data=05%7C01%7Ckzayne%40med.umich.edu%7C8029c56c33f0432f80e708db31f2c4cd%7C1f41d613d3a14ead918d2a25b10de330%7C0%7C0%7C638158690975843407%7CUnknown%7CTWFpbGZsb3d8eyJWIjoiMC4wLjAwMDAiLCJQIjoiV2luMzIiLCJBTiI6Ik1haWwiLCJXVCI6Mn0%3D%7C3000%7C%7C%7C&sdata=69H4yhrgkeHw8xzd3rFnngivYYI6s5UwPjss7ZE5kmk%3D&reserved=0, or unsubscribehttps://nam02.safelinks.protection.outlook.com/?url=https%3A%2F%2Fgithub.com%2Fnotifications%2Funsubscribe-auth%2FA62UM4H6G4JZ4JDDXW2DBHTW63RSHANCNFSM6AAAAAAWMGJUNM&data=05%7C01%7Ckzayne%40med.umich.edu%7C8029c56c33f0432f80e708db31f2c4cd%7C1f41d613d3a14ead918d2a25b10de330%7C0%7C0%7C638158690975843407%7CUnknown%7CTWFpbGZsb3d8eyJWIjoiMC4wLjAwMDAiLCJQIjoiV2luMzIiLCJBTiI6Ik1haWwiLCJXVCI6Mn0%3D%7C3000%7C%7C%7C&sdata=yk%2FB0QzOFqoB9zr5yyE68U2m6Flqyq%2BO2PA4oodOCpY%3D&reserved=0. You are receiving this because you authored the thread.Message ID: @.***>

---

Electronic Mail is not secure, may not be read every day, and should not be used for urgent or sensitive issues

[lmassenet-regad]

Hello, I cannot see what you generated, it is not attached to the previous comment, can you upload the results ?

I think I partially figured out what is happening. This tutorial has been created with an oldest version of the ligand/receptor database, so the plots may not be up-to-date. This may affect the values of global communication scores (as we take into account more interactions) and distributions, although this vignette is restricted to cytokine communication, which did not evolve much in the last version of the database. But the balloon plot values should not be affected.

[zkatmich]

CAF_barplot.pdf CAF_heatmap_pvalue.pdf CAF_network.pdf CAF_balloonplot.pdf

My apologies, here are the files. Thank you again for your assistance.

[lmassenet-regad]

Thanks. I think there is a problem when running the gene.scaling function. In the tutorial: PC.data=gene.scaling(data = PC.data, n=18, db = transform) and CC.data= gene.scaling(data = data, n=4, db = db2)

Normally, after applying this function, the matrix values should be between 0 and 10 for each gene. Maximum values of each interaction is thus 100 in the balloon plot. Can you check head(PC.data) and head(CC.data) after applying the gene.scaling function ? Or maybe you have an error message or a warning at this step ?

I hope it helps, let me know!

[zkatmich]

That helped, the line got commented out somehow where the scaling was applied, the numbers for the balloon plot are matching now. The bar plot and p-value plots still have some slight discrepancies, but not as major as they were before, likely for the reasons you mentioned above. Thank you.