DIA NN compatibility

[Klemens Fröhlich]

dear msstats team,

I am currently testing the software published last year DIA NN, which gives me the following output (see below or attached tsv)

the fragmention quantitation seems to be reported separately but without the info which ions these are.... so I cannot produce a "clean" 10 column format for analysis, or I might be missing something of course!

Can I still analyse this data somehow with MSStats?

Best, Klemens

File.Name	Protein.Group	Protein.Ids	Protein.Names	Genes	PG.Quantity	PG.Normalised	Genes.Quantity	Genes.Normalised	Genes.MaxLFQ	Genes.MaxLFQ.Unique	Modified.Sequence	Stripped.Sequence	Precursor.Id	Precursor.Charge	Q.Value	Protein.Q.Value	PG.Q.Value	GG.Q.Value	Proteotypic	Precursor.Quantity	Precursor.Normalised	Label.Ratio	RT	RT.Start	RT.Stop	iRT	Predicted.RT	Predicted.iRT	First.Protein.Description	Lib.Q.Value	Ms1.Profile.Corr	Ms1.Area	Evidence	CScore	Decoy.Evidence	Decoy.CScore	Fragment.Quant.Raw	Fragment.Quant.Corrected	Fragment.Correlations	MS2.Scan
N:\MFA\MFA654-665_HEK_Ecoli_DIA_with_iRT\MFA654A.mzML	P55011	P55011	S12A2_HUMAN	SLC12A2	282504	286084	282504	286084	281322	281322	AAAAAAAAAAAAAAAGAGAGAK	AAAAAAAAAAAAAAAGAGAGAK	AAAAAAAAAAAAAAAGAGAGAK3	3	0.000256954	0.000455996	0.000415282	0.000416146	1	282504	286084	0	470.075	463.992	475.178	467.287	471.841	465.636	Solute carrier family 12 member 2	0.00293904	0.714551	469640	283.739	0.996095	196.418	0.432299	148771;177351;97033.6;79471.5;61490.9;20711.2;	93985.1;91485.5;97033.6;21597.1;26216.2;0;	0.398117;0.499402;0.962261;0.275779;0.0164255;0.3561;	47339
N:\MFA\MFA654-665_HEK_Ecoli_DIA_with_iRT\MFA654B.mzML	P55011	P55011	S12A2_HUMAN	SLC12A2	226368	191725	226368	191725	339003	339003	AAAAAAAAAAAAAAAGAGAGAK	AAAAAAAAAAAAAAAGAGAGAK	AAAAAAAAAAAAAAAGAGAGAK3	3	0.00310668	0.000392465	0.000358551	0.000359583	1	61822.8	89437.2	0	464.929	461.946	46.9	467.287	471.444	461.051	Solute carrier family 12 member 2	0.00293904	0.562793	156150	274.787	0.856456	205.955	0.29705	23905;54934;43340.2;27749.3;45590.4;18059.1;	7064.6;11418;43340.2;0;0;10600.4;	0.173492;0.27878;0.877492;-0.25966;0.365566;0.389279;	46676
N:\MFA\MFA654-665_HEK_Ecoli_DIA_with_iRT\MFA655A.mzML	P55011	P55011	S12A2_HUMAN	SLC12A2	129249	225992	129249	225992	295599	295599	AAAAAAAAAAAAAAAGAGAGAK	AAAAAAAAAAAAAAAGAGAGAK	AAAAAAAAAAAAAAAGAGAGAK3	3	0.000117041	0.0003885	0.000356379	0.00035727	1	129249	225992	0	467.287	464.264	470.914	467.287	471.423	463.335	Solute carrier family 12 member 2	0.00293904	0.895236	289665	480.766	0.996374	288.883	0.41737	66757.5;56977;45692.5;31070.5;34237.7;24600.9;	41328.3;42227.7;45692.5;17194;20302.5;11549.1;	0.579655;0.778587;0.936687;0.67858;0.539728;0.595337;	46523

[froehlic...@gmail.com]

So I just set every quantitation to the transition y5 and proceeded with the analysis of a benchmark dataset.

the overall analysis worked well, but there were some really crazy outliers in the data ( log2FC >100 ).

Appreciate any input on the matter!

Best, Klemens

[froehlic...@gmail.com]

Okay, I finally found the time to look closer at DIA NN and msstats analysis. 

My mistake was to take the wrong quantitation column which resulted in this weird outlier problem.

However, the main problem still consists: I do have the fragment quant but I dont know which ions were quantified

DIA NN however seems to report the same number of ions for every precursor even if the quant for one transition is empty

7

So I split up the quant info in extra columns and named these columns with y5 to y10

Even though the fragment ion annotation is artificial, the corresponding ions are always annotated the same way.

basically: 
DIA NN output

Exp     Pep    Frag.Quant
Run1  pep1  10,15
Run2  pep1  20,30

becomes

Exp     Pep    y5  y6

Run1  pep1  10  15

Run2  pep1  20  30  

I will post how the analysis will go if anybody else wants to use DIA NN and msstats.

Best, Klemens

[Rayner Queiroz]

Dear Klemens,

I am trying to use DIA-NN myself now. Would you mind sharing your script with me?

best

Rayner

[Mateusz Staniak]

Hi,

I plan to add a converter for DIA-NN output to MSstats. I don't have column definitions, yet, do you know where can I find them? Which column do you use for quantification?

Kind regards,

Mateusz

[Miguel Cosenza]

Hello,

I sent you both some answers directly via email but I actually wanted to share them in this forum thread so other people can look at it if they find it interesting:

Hello Rayner,

This function/script in the link is a modified version of the script from Klemens.

https://github.com/MiguelCos/MSstats_labelfree_preprocessing/blob/master/R/diann2msstats.R

It uses Genes.MaxQ.LFQ as the quantitative variable per feature.


Hello Mateusz,

I believe the documentation of DIA-NN has this information.

https://github.com/vdemichev/DiaNN/blob/master/DIA-NN%20GUI%20manual.pdf

In our lab, we are using the columns labeled as "Genes.MaxLFQ" which have the MaxLFQ quantitation info per feature.

Best wishes,
Miguel

[froehlic...@gmail.com]

Hi

as miguel just posted the documentation can be found on github

all the PG.Quant and LFQs are on protein level.

Precursor level evidence seems to be:

Precursor.Quantity
Precursor.Normalised

Ms1.Area

fragment level quant seems to be:

Fragment.Quant.Raw

Fragment.Quant.Corrected

the precursor quantity does seem to be the sum of all fragment raws...... 

If you want to do protein summarization in msstats and not directly use DIA-NN's protein output, then pragmatically the Precursor.Quant should be correct?

Vadim (the main developer of DIA-NN) answered to me on github within minutes last time I had a question.

Best, Klemens

[Fernando Tobias]

Hello-

I'm looking for a way to use MSStats for analysis DIA-NN outputs. 

I haven't been able to read get the DIA-NN to MSStats preprocessing script to work. Is the input for the scrip the report.tsv from DIA-NN?

Thanks,

Fernando

[Mateusz Staniak]

Hi,

what's the problem with the script?

DIA-NN converter is high on my priorities list, but unfortunately I didn't have a chance to write it, yet

Best,
Mateusz

[Fernando Tobias]

Hello Mateusz,

I attempted to use the function that Miguel shared: 

https://github.com/MiguelCos/MSstats_labelfree_preprocessing/blob/master/R/diann2msstats.R

I'm not quite sure if the function can directly read the Main output report that DIA-NN creates. You will have to excuse me, but my R skills are quite limited. 

I tried running the function after defining diann_data and annotation_file.

> diann_data <- read.delim(file = "C:/DIA-NN/1.8/report_2.tsv")

> annotation_file <- read.csv(file = "C:/DIA-NN/1.8/lynch_diapasef_annotation.csv")

> DIANN_to_MSstats()

Error in mutate(diann_data, File.Name = str_replace(diann_data[[1]], ".*\\\\",  : 

  argument "diann_data" is missing, with no default

Happy to take any advice on converting the report at this time.

Fernando

[Sam Siljee]

Hi Miguel,

I had a look at the code you mentioned, https://github.com/MiguelCos/MSstats_labelfree_preprocessing/blob/master/R/diann2msstats.R

With your permission, I'd like to adapt it for use in my own analysis program.

Best regards,
Sam

[Mateusz Staniak]

Hi,

there is a DIA-NN in MSstats release version now

Kind regards,
Mateusz Staniak

[Sam Siljee]

Thank you Mateusz,

This is very helpful! Unfortunately so far I've only managed to return a dataframe with no rows, but I'll keep trying.

Sam

[Mateusz Staniak]

Hi,

are you using the latest versions of MSstats and MSstatsConvert? At one point there was a bug in the converter that caused empty outputs in some cases, but it was fixed. If your problem persists, could you kindly share a reproducible example of your problem?

Kind regards,

Mateusz Staniak

[Sam Siljee]

Quick response!

I was using MSstatsConvert_1.10.0 and MSstats_4.8.0, I've now updated to MSstatsConvert_1.12.0 and MSstats_4.10.0 which has solved the problem!

Thank you so much for the support by the way, one of the many things that makes MSstats so fantastic!

Sam