search

smith-chem-wisc / FlashLFQ

Ultra-fast label-free quantification algorithm for mass-spectrometry proteomics
GNU Lesser General Public License v3.0
19 stars 15 forks source link

Get LFQ Values for fractionated data at protein level #140

Open bressin opened 6 months ago

bressin commented 6 months ago

Hi, I want to extract the protein LFQ intensities for fractionated data. I tried the sum but it appears to be wrong.

This paper (https://www.ncbi.nlm.nih.gov/pmc/articles/PMC7563855/) states "FlashLFQ can normalize fractionated datasets by using a bounded Nelder–Mead optimizer [33] to find a normalization coefficient for each fraction, similar to MaxLFQ."

So I need to apply the normalization coefficient first to each fraction, but where are they listed? I'm using the standard GALAXY output.

Best regards, Annkatrin

Alexander-Sol commented 6 months ago

Hi Annkatrin,

The normalization coefficients are calculated internally when the "Normalize Intensities" option is selected. The reported protein intensities are derived from the normalized peptide intensities. There is no post-processing normalization required

bressin commented 6 months ago

Here are my settings: PsmIdentificationPath = "msms.psmtsv" SpectraFileRepository = "./spectrum_dir" Silent = false OutputPath = "out" Normalize = true PpmTolerance = 10.0 IsotopePpmTolerance = 5.0 Integrate = false NumIsotopesRequired = 2 IdSpecificChargeState = false MaxThreads = -1 MatchBetweenRuns = true MbrRtWindow = 2.5 RequireMsmsIdInCondition = false BayesianProteinQuant = true ProteinQuantBaseCondition = "Day5" McmcSteps = 500 McmcBurninSteps = 1000 UseSharedPeptidesForProteinQuant = false RandomSeed = 0

Hey, here are my settings. So to get the intensity for a sample with fractions I simply sum all intensity values?

Alexander-Sol commented 6 months ago

Could you share your experimentalDesign.tsv file?

bressin commented 6 months ago

FileName Condition Biorep Fraction Techrep Day5-F-G_SCX_X Day5 3 4 1 Day5-F-G_SCX_5 Day5 3 3 1 Day5-F-G_SCX_4 Day5 3 2 1 Day5-F-G_SCX_3 Day5 3 1 1 Day5-F-G_1D Day5 3 5 1 Day5-D-E_SCX_X Day5 2 4 1 Day5-D-E_SCX_5 Day5 2 3 1 Day5-D-E_SCX_4 Day5 2 2 1 Day5-D-E_SCX_3 Day5 2 1 1 Day5-D-E_1D Day5 2 5 1 Day5-B-C_SCX_3 Day5 1 5 1 Day5-B-C_SCX_X Day5 1 4 1 Day5-B-C_SCX_5 Day5 1 3 1 Day5-B-C_SCX_4 Day5 1 2 1 Day5-B-C_1D Day5 1 1 1 Day4-D_SCX_X Day4 3 5 1 Day4-D_SCX_5 Day4 3 4 1 Day4-D_SCX_4 Day4 3 3 1 Day4-D_SCX_3 Day4 3 2 1 Day4-D_1D Day4 3 1 1 Day4-C_SCX_X Day4 2 5 1 Day4-C_SCX_5 Day4 2 4 1 Day4-C_SCX_4 Day4 2 3 1 Day4-C_SCX_3 Day4 2 2 1 Day4-C_1D Day4 2 1 1 Day4-B_SCX_X Day4 1 5 1 Day4-B_SCX_5 Day4 1 4 1 Day4-B_SCX_4 Day4 1 3 1 Day4-B_SCX_3 Day4 1 2 1 Day4-B_1D Day4 1 1 1 Day3-D_SCX_X Day3 3 5 1 Day3-D_SCX_5 Day3 3 4 1 Day3-D_SCX_4 Day3 3 3 1 Day3-D_SCX_3 Day3 3 2 1 Day3-D_1D Day3 3 1 1 Day3-C_SCX_X Day3 2 5 1 Day3-C_SCX_5 Day3 2 4 1 Day3-C_SCX_4 Day3 2 3 1 Day3-C_SCX_3 Day3 2 2 1 Day3-C_1D Day3 2 1 1 Day3-B_SCX_X Day3 1 5 1 Day3-B_SCX_5 Day3 1 4 1 Day3-B_SCX_4 Day3 1 3 1 Day3-B_SCX_3 Day3 1 2 1 Day3-B_1D Day3 1 1 1 Day2-D_SCX_X Day2 3 5 1 Day2-D_SCX_5 Day2 3 4 1 Day2-D_SCX_4 Day2 3 3 1 Day2-D_SCX_3 Day2 3 2 1 Day2-D_1D Day2 3 1 1 Day2-C_SCX_X Day2 2 5 1 Day2-C_SCX_5 Day2 2 4 1 Day2-C_SCX_4 Day2 2 3 1 Day2-C_SCX_3 Day2 2 2 1 Day2-C_1D Day2 2 1 1 Day2-B_SCX_X Day2 1 5 1 Day2-B_SCX_5 Day2 1 4 1 Day2-B_SCX_4 Day2 1 3 1 Day2-B_SCX_3 Day2 1 2 1 Day2-B_1D Day2 1 1 1 Day1-D_SCX_X Day1 3 5 1 Day1-D_SCX_5 Day1 3 4 1 Day1-D_SCX_4 Day1 3 3 1 Day1-D_SCX_3 Day1 3 2 1 Day1-D_1D Day1 3 1 1 Day1-C_SCX_X Day1 2 5 1 Day1-C_SCX_5 Day1 2 4 1 Day1-C_SCX_4 Day1 2 3 1 Day1-C_SCX_3 Day1 2 2 1 Day1-C_1D Day1 2 1 1 Day1-B_SCX_X Day1 1 5 1 Day1-B_SCX_5 Day1 1 4 1 Day1-B_SCX_4 Day1 1 3 1 Day1-B_SCX_3 Day1 1 2 1 Day1-B_1D Day1 1 1 1 Day0-D_SCX_X Day0 3 5 1 Day0-D_SCX_5 Day0 3 4 1 Day0-D_SCX_4 Day0 3 3 1 Day0-D_SCX_3 Day0 3 2 1 Day0-D_1D Day0 3 1 1 Day0-C_SCX_X Day0 2 5 1 Day0-C_SCX_5 Day0 2 4 1 Day0-C_SCX_4 Day0 2 3 1 Day0-C_SCX_3 Day0 2 2 1 Day0-C_1D Day0 2 1 1 Day0-B_SCX_X Day0 1 5 1 Day0-B_SCX_5 Day0 1 4 1 Day0-B_SCX_4 Day0 1 3 1 Day0-B_SCX_3 Day0 1 2 1 Day0-B_1D Day0 1 1 1

Alexander-Sol commented 5 months ago

Thanks for providing that. What make you say that the values appear to be wrong?

Would it be useful if we provided the normalization coefficients as an output?

bressin commented 5 months ago

If the intensities are already weighted I don't need the coefficients, but I think it would be still nice to report them somewhere. However, sometimes the summed LFQ values don't match the log2FC provided. Maybe it is worth checking.