write_adat function not working, even on example data that is unaltered.

[aabarrera2]

Hi Stu,

The write_adat function gives me this message , with my own data and with example data - ADAT passed checks and trapsx Attributes has only 3 entries: 'names', 'row.names', 'class'Error in mutate.soma_adat(x, blank_col = NA_character_) :
is.intact.attributes(.data) is not TRUE

whats strange is that ive called both 'is.intact.attributes' and 'is.soma_adat()'. and both are true. I know these are both requirements for write_adat.

I am just wondering if you have an idea why.
I am using data from assay version 3.2 and the !version of my file from somalogic is 1.2

[stufield]

Hi @aabarrera2 , this seems a bit like a breaking change from dplyr::mutate() v1.0.8. Can you tell me which version of dplyr you are using and which version of SomaDataIO?
The most recent version of SomaDataIO (v5.3.0) has a bugifx for this specific issue. So updating (if you're not on it already) will likely solve this.
The reason your attributes are intact outside the write_adat() call is that they are initially ok, but internally mutate() breaks them, then you get the error that they're broken (attributes are required to regenerate the Header information of the ADAT when you write it to file).
BTW: the new SomaDataIO v5.3.0 doesn't even use dplyr::mutate() anymore, so this issue will simply go away, even without the bugfix.

[aabarrera2]

Hi Stu, Thank you so much for a swift reply to my issue. I thought i was using the recent version of somadataIO but i wasnt. Updating it helped me write an adat file successfully. However, now im having trouble reading an adat file that ive written. (first I read in my somascan adat, joined it with my metadata frame, then wrote the adat. Reading this new adat file has not been successful (I was able to do this successfully a few months ago, but running the codes again give me Error in scan(file = file, what = what, sep = sep, quote = quote, dec = dec, : scan() expected 'a real', got 'TargetFullName' Is this something youve noticed others having issues with too? I wonder if you have an idea why read.adat() isnt able to read in an adat file i created. Albert

…

On Fri, Apr 15, 2022 at 12:59 PM Stu Field ***@***.***> wrote: Hi @aabarrera2 <https://github.com/aabarrera2> , this seems a bit like a breaking change from dplyr::mutate() v1.0.8. Can you tell me which version of dplyr you are using and which version of SomaDataIO? The most recent version of SomaDataIO (v5.3.0) has a bugifx for this specific issue. So updating (if you're not on it already) will likely solve this. The reason your attributes are intact *outside* the write_adat() call is that they are initially ok, but internally mutate() breaks them, then you get the error that they're broken (attributes are required to regenerate the Header information of the ADAT when you write it to file). — Reply to this email directly, view it on GitHub <#18 (comment)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/ANJBPPGCTGEU7U2WU4J3M5LVFGN5NANCNFSM5TQ2M4JQ> . You are receiving this because you were mentioned.Message ID: ***@***.***>

[stufield]

So you're trying to read in an ADAT with v5.3.0 code that was written with < v5.3.0 code? Correct?
I think it would be best to re-create the joined ADAT from scratch, with v5.3.0 code so that you're not crossing versions.
Without a proper reprex I'm not sure I'll be able to reproduce the error on my end, especially since we don't use scan() anywhere in SomaDataIO. Are you calling it somewhere and getting this error?

To try: back rev to the version of SomaDataIO used to write the modified ADAT, can read_adat() parse that one?

[stufield]

This reproduces the scan() bug. Likely an alignment issue in write_adat()

example_data |>
  head() |>
  dplyr::select(SlideId, Subarray, Sex, seq.10000.28, seq.10001.7) |> 
  write_adat("out.adat")
foo <- read_adat("out.adat")

[aabarrera2]

Thanks so much for looking into that, Stu. I am new to the bio informatics world and have been tasked with a data analyses of somascan with our meta data and it has been a struggle to do simple analyses like differential expression between 2 groups (my group one has coronary artery calcium, and group 2 has none present). I’ve successfully merged the data frames in such a way that all my rows are different observations and I have over 1.3k columns which are all either clinical variables or protein RFUs. I’m not sure if this is the best way to do the analysis though.it’s a steep learning curve for sure but really exciting!

…

On Fri, Apr 22, 2022 at 15:40 Stu Field ***@***.***> wrote: This reproduces the scan() bug. Likely an alignment issue in write_adat() example_data |> head() |> dplyr::select(SlideId, Subarray, Sex, seq.10000.28, seq.10001.7) |> write_adat("out.adat")foo <- read_adat("out.adat") — Reply to this email directly, view it on GitHub <#18 (comment)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/ANJBPPCYHQHYOKUZRGBLMN3VGL6BTANCNFSM5TQ2M4JQ> . You are receiving this because you were mentioned.Message ID: ***@***.***>

[stufield]

@aabarrera2 , there's a nice example in the README for how we suggest performing a simple 2-group analysis. Might be a good resource for you to follow as a template for your data. See the section on 2 group t-test analyses.

[KirkDSL]

If your analysis is a simple 2-group analysis (e.g., t-test), you may be able to use our Statistics and Data Visualization software here: https://somalogic.com/statistical-analysis-tool/

Here's the link to the User's Guide describing what functionality the tool provides: https://stats.somalogic.com/userguide/

[aabarrera2]

Thank you Stu, I actually successfully used the proviz tool and was able to output t tests and correlations. It’sa great start but my boss wants me to go further and adjust for confounding variables with regressions and plot heat maps of the fold changes. Is there any kind of documents for those more in depth analyses?

…

On Fri, Apr 22, 2022 at 15:55 Stu Field ***@***.***> wrote: @aabarrera2 <https://github.com/aabarrera2> , there's a nice example in the README for how we suggest performing a simple 2-group analysis. Might be a good resource for you to follow as a template for your data. See the section on 2 group t-test analyses. — Reply to this email directly, view it on GitHub <#18 (comment)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/ANJBPPG6PS3H2ZITQ6E7E5DVGL74FANCNFSM5TQ2M4JQ> . You are receiving this because you were mentioned.Message ID: ***@***.***>

[KirkDSL]

I'm glad you were able to use the tools successfully! Unfortunately, simple analyses is where it stops. More sophisticated analyses definitely require writing code, as you are doing now. Hopefully we'll be able to add more functionality to the tool.

[aabarrera2]

Hi Stu, Ive recreated the ADAT file using the latest write_adat, and it says all checks and traps passed, but when I read the file back in, i get this message: Error in scan(file = file, what = what, sep = sep, quote = quote, dec = dec, : scan() expected 'a real', got 'FEMALE' When i add debug and verbose = TRUE, I obtain this: Error: The Col.Meta shift 34 does not match the length stated in ^ROW_DATA row 171 -- visually inspect ADAT I get this error even after simply reading in the raw adat from somalogic, and then writing it back into adat, and then reading it back gives me this error. Ive updated all my packages with tidyverse.

somascan <-

read_adat('~/Desktop/R_WD_2/SomaLogicData/CHI-20-017_v3.2_Serum.20210415.adat')

write_adat(somascan,

"~/Desktop/R_WD_2/SomaLogicData/soma_step4_samples.adat") ✓ ADAT passed all checks and traps.✓ ADAT written to: '~/Desktop/R_WD_2/SomaLogicData/soma_step4_samples.adat'

n <- read_adat("~/Desktop/R_WD_2/SomaLogicData/soma_step4_samples.adat")

Error in scan(file = file, what = what, sep = sep, quote = quote, dec = dec, : scan() expected 'a real', got 'TargetFullName' Is this an error anyone else has posited? Albert

…

On Mon, Apr 18, 2022 at 7:22 PM Stu Field ***@***.***> wrote: So you're trying to read in an ADAT with v5.3.0 code that was written with < v5.3.0 code? Correct? I think it would be best to re-create the joined ADAT from scratch, with v5.3.0 code so that you're not crossing versions. Without a proper reprex I'm not sure I'll be able to reproduce the error on my end, especially since we don't use scan() anywhere in SomaDataIO. Are you calling it somewhere and getting this error? — Reply to this email directly, view it on GitHub <#18 (comment)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/ANJBPPGT3Z4O2BZCNYFNJW3VFXVCHANCNFSM5TQ2M4JQ> . You are receiving this because you were mentioned.Message ID: ***@***.***>

[stufield]

Hi @aabarrera2 ,

This appears to be the same error as the original post correct?
Can you confirm which version of SomaDataIO you are using to generate this error?
SomaDataIO v5.3.0 should contain the bug, but it was fixed (I believe) as a result of this issue, so it should be fixed in the current HEAD of the GitHub repository (i.e. the development version).
As an aside, I am currently working on freezing the package where it stands into a bugfix v5.3.1 version to incorporate this bugfix into a frozen version.

If you could confirm the development fixes this issue I would appreciate it.
Thanks!

Stu

[aabarrera2]

Hi Stu We can close this ticket. All is well.

…

On Tue, Oct 11, 2022 at 10:54 Stu Field ***@***.***> wrote: Hi @aabarrera2 <https://github.com/aabarrera2> , This appears to be the same error as the original post correct? Can you confirm which version of SomaDataIO you are using to generate this error? SomaDataIO v5.3.0 should contain the bug, but it was fixed (I believe) as a result of this issue, so it should be fixed in the current HEAD of the GitHub repository (i.e. the development version). As an aside, I am currently working on freezing the package where it stands into a bugfix v5.3.1 version to incorporate this bugfix into a frozen version. If you could confirm the development fixes this issue I would appreciate it. Thanks! Stu — Reply to this email directly, view it on GitHub <#18 (comment)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/ANJBPPDUXIWS7GOYTUA6M5TWCV5SZANCNFSM5TQ2M4JQ> . You are receiving this because you were mentioned.Message ID: ***@***.***>

[stufield]

I believe it automatically closed when the bugfix was merged. All good.