Sign in

Fengchao

@fcyucn.bsky.social
1K followers 293 following 40 posts

Research Investigator at U of M. Interested in proteomics, etc. Developer of FragPipe, MSFragger, IonQuant, etc.

PostsRepliesMedia
Reposted by Fengchao
Keri Backus @keribackus.bsky.social · 28/04/2026
Big congrats to first authors Chau Ngo and Sho Takechi and a big thank you to all the authors and especially to our collaborators including @nesvilab.bsky.social @fcyucn.bsky.social @marcliesa.bsky.social and Ajit Divakaruni
121
Reposted by Fengchao
Ludwig Lautenbacher @llautenbacher.bsky.social · 11/11/2025
Exited to share our latest work! Out now in @natcomms.nature.com Koina aims to transform how #proteomics uses machine learning. You no longer need to be a tech wizard to use ML and now can easily run #ML models. Integrated with FragPipe, Skyline and EncyclopeDIA! www.nature.com/articles/s41...
nature.com
Koina: Democratizing machine learning for proteomics research - Nature Communications
Koina is an open-source, online platform that simplifies access to machine learning models in proteomics, enabling easier integration into analysis tools and helping researchers adopt and reuse ML mod...
04116
Fengchao @fcyucn.bsky.social · 30/10/2025
This project started 5 years ago. It led us to add isotope-labeling support to #FragPipe/#IonQuant. Since then, the tools have grown so much and are now widely used in #Chemoproteomics. Huge thanks to everyone, and special thanks to @stephanhacker2.bsky.social and @pzanon.bsky.social
0145
Reposted by Fengchao
Stephan Hacker @stephanhacker2.bsky.social · 30/10/2025
How can we study target engagement and selectivity of covalent inhibitors? Which electrophilic probes are best suited to study a certain amino acid? Our study on "Profiling the proteome-wide selectivity of diverse electrophiles" is published in Nature Chemistry.(1/7) www.nature.com/articles/s41...
nature.com
Profiling the proteome-wide selectivity of diverse electrophiles - Nature Chemistry
Covalent inhibitors are powerful entities in drug discovery. Now the amino acid selectivity and reactivity of a diverse electrophile library have been assessed proteome-wide using an unbiased workflow...
28734
Reposted by Fengchao
Michael MacCoss @maccoss.bsky.social · 15/09/2025
Proteomics Webinar: DIA with FragPipe, DIA-NN, and Skyline Presenters: Eduard Sabidó and Brendan MacLean When: Tuesday, September 16, 8am (Pacific Time) Register Now ... skyline.ms/project/home... #massspec #proteomics
skyline.ms
Start Page: /home/software/Skyline/events/2025 Webinars/Webinar 26
03216
Fengchao @fcyucn.bsky.social · 10/09/2025
It was a great pleasure to teach #FragPipe at the Biological Proteomics for Beginners workshop at #UCSD, sponsored by Thermo Fisher Scientific. We had a fantastic group of grad students, postdocs, and professors. Yes, I even got to teach UCSD professors how to analyze bottom-up proteomics data 😁
1181
Reposted by Fengchao
Alexey Nesvizhskii @nesvilab.bsky.social · 20/08/2025
#MSFragger Open Search has been around for a while now and used by mass spec folks to screen for chemical artifacts and adducts, e.g. in chemoproteomics data. Happy to see it got 'discovered' by a broader community who are now reporting all sort of cool biological PTMs www.nature.com/articles/s41...
nature.com
Nucleoside diphosphate kinase A (NME1) catalyses its own oligophosphorylation - Nature Chemistry
Our understanding of how post-translational modification—protein phosphorylation—impacts the complexity of eukaryotic signalling pathways is continuously expanding. Now, protein oligophosphorylation h...
1291
Fengchao @fcyucn.bsky.social · 14/08/2025
Interesting idea. May I ask two questions? 1. Is the precursor detection performed before or after the database search? 2. If it is performed before the search, how can one identify the complementary fragment ions without knowing the precursor mass or peptide sequence? Thanks
100
Reposted by Fengchao
Stephan Hacker @stephanhacker2.bsky.social · 12/08/2025
Interested in the proteome-wide selectivity of diverse electrophiles? The proteomics data for our study on this headed by @pzanon.bsky.social are now public on @pride-ebi.bsky.social: www.ebi.ac.uk/pride/archiv... Full story: chemrxiv.org/engage/chemr... #ChemBio #ChemSky #ChemicalProteomics
chemrxiv.org
Profiling the proteome-wide selectivity of diverse electrophiles
Targeted covalent inhibitors are powerful entities in drug discovery, but their application has so far mainly been limited to addressing cysteine residues. The development of cysteine-directed covalen...
0218
Reposted by Fengchao
Alexey Nesvizhskii @nesvilab.bsky.social · 01/08/2025
Conventional proteomics searches struggle with many modifications and fully open searches may be difficult to interpret. We introduce a "detailed" mass offset search in #MSFragger boosting interpretability and localization especially in complex cases (e.g. FPOP data): www.biorxiv.org/content/10.1...
0346
Reposted by Fengchao
Nicholas McCurtin, PhD 🧫 🏳️‍🌈 @nicholasmccurtin.bsky.social · 11/07/2025
Don't tell PD, but I have quietly switched all of my analyses to Fragpipe. What an extremely powerful software. I'm often amazed at the sheer quantity of information I can get using Fragpipe. Thanks to everyone who recommended it.
2122
Fengchao @fcyucn.bsky.social · 15/07/2025
Excited to be one of the guest editors! We’re calling for papers on cutting-edge #proteomics and #metabolomics in #multiomics research. Looking forward to your submissions!
0153
Fengchao @fcyucn.bsky.social · 30/05/2025
In addition to these exciting events, I’ll also be one of the panel members at the single-cell proteomics Evening Workshop on Wednesday. Looking forward to the discussions ahead!
071
Reposted by Fengchao
Alexey Nesvizhskii @nesvilab.bsky.social · 30/05/2025
Wow, a record breaking number of the Nesvizhskii lab members attending #ASMS2025! 9 posters, 3 evening workshops, and one Bioinformatics Hub on #FragPipe. Plus multiple collaborative posters with other groups. See you in Baltimore! PS. Below is our recent group photo, including all those attending
0358
Reposted by Fengchao
Alexey Nesvizhskii @nesvilab.bsky.social · 23/05/2025
The power of #MSFragger open search! “we applied the mass-tolerant search engine MSfragger and found that phosphorylation as well as ubiquitination were well preserved after XDNAX. To our great interest, we found an additional modification of 321 Da occurring only in the irradiated SILAC channel”
0142
Reposted by Fengchao
Alexey Nesvizhskii @nesvilab.bsky.social · 15/05/2025
If there any #Sciex decision makers here on Bluesky - I urge you to reconsider. Skyline/Proteowizard support is not only important for your customers using these tools, but it also benefits other bioinformatics efforts that depend on these tools.
0259
Fengchao @fcyucn.bsky.social · 08/05/2025
We didn’t cherry-pick the dataset for the benchmark—FragPipe also delivers excellent performance on another dataset (PXD028735). Questions are always welcome! We're always happy to hear your feedback. (3/3)
010
Fengchao @fcyucn.bsky.social · 08/05/2025
Peptide-level LFQbench-style plot using the PXD003881 (IonStar) dataset. (2/3)
110
Fengchao @fcyucn.bsky.social · 08/05/2025
We also benchmarked LFQ precision and accuracy against the latest versions of other popular tools. Here's an protein-level LFQbench-style plot using the PXD003881 (IonStar) dataset. (1/3)
110
Fengchao @fcyucn.bsky.social · 08/05/2025
In this release, one of the major improvements is LFQ using IonQuant. Thanks to this excellent preprint (www.biorxiv.org/content/10.1...), we identified and fixed a suboptimal step in the XIC. We're always eager to listen to feedback from the community!
1254
Reposted by Fengchao
Qian Cheng @qchengx1e3.bsky.social · 15/04/2025
MSFragger-DDA+ is a database search tool that enhances peptide identification by searching the full isolation window. #Proteomics @fcyucn.bsky.social @nesvilab.bsky.social www.nature.com/articles/s41...
nature.com
MSFragger-DDA+ enhances peptide identification sensitivity with full isolation window search - Nature Communications
Proteomics often misses co-fragmented peptides in DDA data. Here, the authors introduce MSFragger-DDA+, a database search tool that enhances peptide identification by searching the full isolation wind...
081
Reposted by Fengchao
Alexey Nesvizhskii @nesvilab.bsky.social · 09/04/2025
DDA is still great for many applications, and #MSFragger-DDA+ improves peptide identification sensitivity via full isolation window search. Huge boosts in IDs, including Astral DDA! Fully integrated in #FragPipe, simply annotate your DDA files as DDA+ and RUN. www.nature.com/articles/s41...
nature.com
MSFragger-DDA+ enhances peptide identification sensitivity with full isolation window search - Nature Communications
Proteomics often misses co-fragmented peptides in DDA data. Here, the authors introduce MSFragger-DDA+, a database search tool that enhances peptide identification by searching the full isolation wind...
13813
Fengchao @fcyucn.bsky.social · 04/03/2025
Feature request well received🫡
050
Reposted by Fengchao
Alexey Nesvizhskii @nesvilab.bsky.social · 04/03/2025
One of the most significant and challenging projects of my career so far. PepCentric: a scalable computational platform utilizing novel 2-D fragment indexing for rapid peptide-centric searches, enabling proteogenomics searches against billions of spectra in seconds. www.biorxiv.org/content/10.1...
37323
Fengchao @fcyucn.bsky.social · 04/01/2025
It doesn't "predict" the library. It perform the spectrum-centric search to get the IDs and build the experimental library. And for the DIA without ion mobility, it is basically what DIA-Umpire is doing: www.nature.com/articles/nme...
nature.com
DIA-Umpire: comprehensive computational framework for data-independent acquisition proteomics - Nature Methods
The computational workflow of DIA-Umpire allows untargeted peptide identificationdirectly from DIA (data-independent acquisition) proteomics data without dependence on a spectral library for data extr...
110
Reposted by Fengchao
Alexey Nesvizhskii @nesvilab.bsky.social · 04/01/2025
Sitting at Pike Place Market in Seattle at 7am in the morning, sipping coffee with my son, a long layover on the way home from Taipei. Otherwise I would do a long post. DiaTracer in #FragPipe work really well, making library-free analysis of any diaPASEF data possible. www.nature.com/articles/s41...
nature.com
diaTracer enables spectrum-centric analysis of diaPASEF proteomics data - Nature Communications
Data-independent acquisition advances proteomics quantification. Here, the authors present diaTracer, a spectrum-centric tool for diaPASEF data that supports broad proteomics applications, enabling di...
0377
Fengchao @fcyucn.bsky.social · 04/01/2025
Perfect timing: www.nature.com/articles/s41...
nature.com
diaTracer enables spectrum-centric analysis of diaPASEF proteomics data - Nature Communications
Data-independent acquisition advances proteomics quantification. Here, the authors present diaTracer, a spectrum-centric tool for diaPASEF data that supports broad proteomics applications, enabling di...
130
Reposted by Fengchao
PastelBio @pastelbio.bsky.social · 03/01/2025
diaTracer enables spectrum-centric analysis of diaPASEF proteomics data www.nature.com/artic... --- #proteomics #prot-paper
191
Fengchao @fcyucn.bsky.social · 04/01/2025
Mostly, yes, AFAIK. The Gaussian feature detection was also inspired by Alexey's previous work doi.org/10.1021/acs.... Although that work was for top-down but we found that the ion mobility shapes were similar to timsTOF's.
doi.org
IMTBX and Grppr: Software for Top-Down Proteomics Utilizing Ion Mobility-Mass Spectrometry
Top-down proteomics has emerged as a transformative method for the analysis of protein sequence and post-translational modifications (PTMs). Top-down experiments have historically been performed prima...
020
Reposted by Fengchao
Alexey Nesvizhskii @nesvilab.bsky.social · 10/12/2024
Looks interesting, and they already coupled it with MSFragger/FragPipe to search MS data using their fasta. I recommend using the group FDR option (available in FragPipe; requires annotating the headers in fasta in certain way) or a two-pass search option (canonical DB->unassigned mzML->custom DB).
181
Fengchao @fcyucn.bsky.social · 06/12/2024
No, we didn't exclude those shifts when looking for decoys. As you can see, there is room for us to improve. And I would like to do so if time permits.
010
Fengchao @fcyucn.bsky.social · 06/12/2024
Credit goes to the giants: the PNNL lab, the SEQUEST family of developers, Matthew, Lukas, and many others. As an old guy said "As always, we are standing on the shoulders of giants. The proteomics community is so amazing because we build on each other's success."
010
Fengchao @fcyucn.bsky.social · 06/12/2024
Yes, I agree with it. I think I have responded and "liked" all your comments. Let me know if I missed something or said something wrong/nonsense. Happy to discuss and correct myself.
030
Fengchao @fcyucn.bsky.social · 06/12/2024
It reminds me that we also thought about using the decoys from a region totally different from target is also not good. (Sorry that I didn't recall the whole story from years ago. I might still miss something...)
110
Fengchao @fcyucn.bsky.social · 06/12/2024
pubs.acs.org/doi/10.1007/... and pubs.acs.org/doi/10.1021/... discussed it in depth. Basically, there is a 1.0005 period in the mass axis. The decoy's shift better to be N*1.0005 so that it has false peptide but similar property as the target.
330
Fengchao @fcyucn.bsky.social · 06/12/2024
Ah, sorry, I pasted the wrong figures in a post I deleted. Let me redo it. There are two types of the works discussing about 1.0005 Th shift. One is from The and Kall 2020 (www.nature.com/articles/s41...). See the figure. Another one is from the lab in PNNL and SEQUEST family. Let me find a paper
110
Fengchao @fcyucn.bsky.social · 06/12/2024
Yes, exactly! That is why we don't rely on that assumption and shift the m/z instead. As you agree that different peptides won't have different MS1 "profile", the Uri's null assumption "...is a false, random one" might not be true because you don't know if shifting RT gives you false peptide.
120
Fengchao @fcyucn.bsky.social · 06/12/2024
I agree with what the second paragraph says as I have mentioned in my response to another post bsky.app/profile/fcyu...
010
Fengchao @fcyucn.bsky.social · 06/12/2024
That's the reason that we shifted the m/z. There are literatures saying that with N*1.0005 Th shift, there is a low chance to have a real peptide in that location (correct me if I am wrong). We don't assume that different peptides have different MS1 "profile".
230
Fengchao @fcyucn.bsky.social · 06/12/2024
If the above assumption doesn't exist, how can you define "false" in the "the considered match between the acceptor MS1 peak/feature with the donor peptide is a false, random one", and how to guarantee that it is false.
210
Fengchao @fcyucn.bsky.social · 06/12/2024
Thanks for sharing. Truly appreciate it. If I understand it correctly, Uri's comment implies that there is an assumption: different peptides have different MS1 "profile" (e.g. XIC shapes, isotope distributions). A randomized RT location might have a real peptide (maybe) with the same charge.
220
Fengchao @fcyucn.bsky.social · 06/12/2024
Right, with nonspecific search, the computer probably need to have a large memory size (64 ~256 GB). You also could enable the "split database" option in the "MSFragger".
220
Fengchao @fcyucn.bsky.social · 06/12/2024
Yes, you can. Load the DIA_SpecLib_Quant_diaPASEF workflow and adjust the enzyme setting.
030
Fengchao @fcyucn.bsky.social · 04/12/2024
Yeah, this is a good point. Then, how do you make sure that the decoy you find is not the same as the target or another real peptide?
130
Fengchao @fcyucn.bsky.social · 04/12/2024
That is one thing I want to figure out by myself or asking you: why do you shift the RT instead of m/z? We followed other giants (see the refs in our paper) by shifting N*1.0005 Th. Did you find shifting RT produces better decoys? Happy to discuss in private if that works better for you.
310
Fengchao @fcyucn.bsky.social · 04/12/2024
I totally agree with everything you've just said. We had considered incorporating peptide identification's posterior error probability into IonQuant's Bayesian model to "penalize" the FDR. Unfortunately, I became sidetracked with other projects and never had the chance to implement the idea.
030
Fengchao @fcyucn.bsky.social · 04/12/2024
Congrats on the excellent work! I just replied to @metamorpheus.bsky.social's post and wanted to share the link bsky.app/profile/fcyu.... Happy to discuss here or via email. Any criticisms or challenges are also welcome. I just want to learn from you all, have no intention to argue with anyone.
120
Fengchao @fcyucn.bsky.social · 04/12/2024
3. The single-cell data is a little "special". I will reproduce the analysis by myself when I have time, but may I ask if you have looked at the single-cell data to see if what you observed is reproducible?
230
Fengchao @fcyucn.bsky.social · 04/12/2024
2. FDR is the expectation of the FDP. Having FDP ("percent false discoveries" in Figure 3) around the FDR is what it is supposed to be. While the manuscript does not explicitly mention PIP-ECHO's FDP values, it appears that both PIP-ECHO and IonQuant exhibit similar deviations around the FDR.
130
Fengchao @fcyucn.bsky.social · 04/12/2024
1. IonQuant's goal is controlling the MBR FDR, not the FDR of peptide identification + MBR. Your Figure 3 has shown that IonQuant's orange bar is lower than 1% and also lower than PIP-ECHO (I will comment on the single-cell data later).
241