anesvi
commented
3 years ago Hi Todd,
-
There is always median centering at the PSM level after taking ratio to the reference channel (or virtual channel, which is an average of all cannels). “PSM norm” refers to more advanced, retention time-based median centering (where median centering is done separately for each RT bin, with the RT range divided into 10 RT bins). We implemented it, and did not see any improvement in most datasets, so left it unchecked. Hui-Yin, have we recently tested this normalization ? Do we describe this option on our website?
-
This is similar to what is done in https://pubs.acs.org/doi/10.1021/acs.jproteome.0c00209. PSMs are sorted by intensity, and the ratio of the ‘median intensity’ PSM is taken as gene/protein ratio (instead of taking median of all ratios from all PSMs for that gene/protein). However, it does not seem to work well except in cases where you have huge fold changes, like datasets with spiked-in proteins giving some very high (like 10:1, 20:1 and higher ratios.
-
If proteins with Empty GN= are not in the Protein-level file, that seems like a bug. Hui-Yin, can you take a look? Thanks for using the tools, Todd!
Best Alexey
From: enonimos @.> Sent: Monday, March 22, 2021 2:32 PM To: Nesvilab/TMT-Integrator @.> Cc: Subscribed @.***> Subject: [Nesvilab/TMT-Integrator] Advanced Options and TMT report in FragPipe 15 (#15)
External Email - Use Caution
Hi, I am very much enjoying the TMT-Integrator functions in FragPipe and have been testing it out on several datasets. In my testing, a few questions came up.
-
In the TMT-Integrator tab, I see under “Advanced Options” that the PSM norm option is not checked by default. I have read in previous posts, and in Clark et al. 2019, that reference abundance normalization is performed first at the PSM and then again at the gene/protein/etc levels. Is this description of reference channel PSM level normalization the same as the “PSM norm” option in FragPipe, or is "PSM norm" an additional normalization step?
-
In FragPipe v15.0, I see there is a new “weighted” aggregation method. How is this performed and have you found a specific application where this is superior to median aggregation?
-
RE: TMT-report. I have a few proteins that are custom database entries, which did not have gene annotations (e.g. “GN=”) in the FASTA header. These proteins are found in the combined_proteins.tsv file, but were excluded from the TMT report. I could imagine they would be excluded if I selected “Gene” summarization, but I had selected “Protein” summarization. After discovering this result, I modified my custom sequences to specify a "gene" in the header file and this solved the issue. However, this may be a broader issue as in human and mouse UniProt (reviewed) protein sequence databases there are a few hundred protein sequences that do not have an annotated gene.
Thanks again for writing this great tool, the different normalization and summarization options are very useful, and integration within the FragPipe GUI makes it easy to analyze large datasets quickly. Best, Todd
— You are receiving this because you are subscribed to this thread. Reply to this email directly, view it on GitHubhttps://github.com/Nesvilab/TMT-Integrator/issues/15, or unsubscribehttps://github.com/notifications/unsubscribe-auth/AIIMM62SJ5IC5BKOVQHLGU3TE6ELHANCNFSM4ZTVBAIQ.
Electronic Mail is not secure, may not be read every day, and should not be used for urgent or sensitive issues
enonimos
MiguelCos
huiyinc
Hi, I am very much enjoying the TMT-Integrator functions in FragPipe and have been testing it out on several datasets. In my testing, a few questions came up.
In the TMT-Integrator tab, I see under “Advanced Options” that the PSM norm option is not checked by default. I have read in previous posts, and in Clark et al. 2019, that reference abundance normalization is performed first at the PSM and then again at the gene/protein/etc levels. Is this description of reference channel PSM level normalization the same as the “PSM norm” option in FragPipe, or is "PSM norm" an additional normalization step?
In FragPipe v15.0, I see there is a new “weighted” aggregation method. How is this performed and have you found a specific application where this is superior to median aggregation?
RE: TMT-report. I have a few proteins that are custom database entries, which did not have gene annotations (e.g. “GN=”) in the FASTA header. These proteins are found in the combined_proteins.tsv file, but were excluded from the TMT report. I could imagine they would be excluded if I selected “Gene” summarization, but I had selected “Protein” summarization. After discovering this result, I modified my custom sequences to specify a "gene" in the header file and this solved the issue. However, this may be a broader issue as in human and mouse UniProt (reviewed) protein sequence databases there are a few hundred protein sequences that do not have an annotated gene.
Thanks again for writing this great tool, the different normalization and summarization options are very useful, and integration within the FragPipe GUI makes it easy to analyze large datasets quickly. Best, Todd