sgessulat Profile Banner
Siegfried Gessulat Profile
Siegfried Gessulat

@sgessulat

Followers
308
Following
1K
Media
24
Statuses
341

Technical Lead #MachineLearning at @msaid_de

Berlin, Germany
Joined July 2008
Don't wanna be here? Send us removal request.
@MartinFrejno
Martin Frejno
6 months
#TeamMassSpec, don't miss my talk about double dipping in DIA data analysis on Tuesday afternoon at #ASMS2025! It's not as savoury as the one below, but will contain a couple of spicy takes. ๐Ÿ™ƒ
0
1
4
@sgessulat
Siegfried Gessulat
6 months
5/5 All of @msaid_de at #ASMS2025 can be found at https://t.co/CpVNJ2Z32O Two more highlights: - @MartinFrejno's oral TOA pm 04:10 shows a confirmation bias in DIA that inflates data completeness - @dzolg's oral ThOA pm 03:30 shows our platform Enjoy the conference!
Tweet card summary image
msaid.de
MSAID revolutionizes proteomics analysis with AI, providing cutting-edge software and data analysis services using powerful machine learning algorithms.
0
0
0
@sgessulat
Siegfried Gessulat
6 months
4/5 See our newly introduced TimsTOF โœˆ๏ธ support on Tuesday at TP 507. Tobias shows how Chimerys 5 and Inferys 5 perform on dda-PASEF and dia-PASEF and outperform or match alternatives out-of-the-box.
1
0
0
@sgessulat
Siegfried Gessulat
6 months
3/5 What about speed? ๐Ÿƒโ€โ™€๏ธ Tokenization can compress the input by more than 100x. Weight quantization doubles our speed, and knowledge distillation gives another 30%. Combined, Inferys 5 is faster than Inferys 4 at the same high accuracy.
1
0
0
@sgessulat
Siegfried Gessulat
6 months
2/5 Instead of amino acid tokens, we show Inferys 5 the full molecular structure ๐Ÿงฌ. Tokenization and data augmentation enable us to vary the structure's detail level. We can trade speed for out-of-distribution generalization similar to chain-of-thought reasoning models.
1
0
0
@sgessulat
Siegfried Gessulat
6 months
๐Ÿงต1/5 Although I'll miss #ASMS2025 I'm hugely excited about @LMamisashvili's oral ๐Ÿ•ต๏ธโ€โ™€๏ธ TOE am 09:50. Inferys 5 is our biggest ML update in years. We support any PTM input in principle, dramatically increase speed, and add TimsTOF support. All with a single model.
1
0
0
@msaid_de
MSAID
7 months
Join us for our Breakfast Seminar at #ASMS2025 and discover the latest updates to CHIMERYS 5 and the MSAID Platform. Learn about new features and hear real-world applications from fellow scientists who are using our technology in their research. ๐Ÿ”— Register here:
0
1
2
@msaid_de
MSAID
6 months
Only one week left for #ASMS2025! Join us for our talk on accurate and fast zero-shot peptide spectrum prediction, featuring a deep learning approach that predicts fragment ion intensitiesโ€”even for peptides with unseen post-translational modifications. ๐Ÿ“… Tuesday, June 3 |
0
2
3
@msaid_de
MSAID
1 year
2 weeks until #HUPO2024 in Dresden! Join us for an insightful talk on peptide- and spectrum-centric search engines. We will dive into comparisons of various algorithms for peptide identification and showcase AI-driven innovations in bioinformatics. ๐Ÿ“… Tuesday, Oct 22, 2024 |
0
2
9
@sgessulat
Siegfried Gessulat
1 year
I played with a version of Prosit with a 2nd objective: predicting m/z of fragment ions! This feeds m/z info via backpropagation which can improve overall model performance. At inference time, the m/z branch should be dropped ofc. It's a fun trick but unnecessary today.
@MagnusPalmblad
Magnus Palmblad
1 year
@pwilmarth @neely615 @ProteomicsNews I have definitely seen some dubious applications, such as predicting isotopic distributions with deep learning. Maybe predicting masses of peptides with deep learning is would be next? Sounds like a joke, but I bet there are journals in which this could be published...
0
0
3
@MistralAI
Mistral AI
1 year
314
599
5K
@msaid_de
MSAID
1 year
What an incredible company retreat in Brixen, Italy! We enjoyed a week full of scientific and technical discussions, rafting, and fun board games. It was a joy to connect in person, strengthening our bonds and creating unforgettable memories. A big thank you to the entire team
0
3
6
@msaid_de
MSAID
1 year
MSAID is excited to announce its membership in the NVIDIA Inception Program! This milestone propels us further in our quest to transform the field of proteomics with cutting-edge AI and machine-learning innovations. Learn more about the MSAID Platform for processing: The
0
3
11
@msaid_de
MSAID
1 year
#IKOM2024 Join us today at Booth 14! Discover exciting career opportunities and learn more about our open positions. We look forward to meeting you! #Hiring #SoftwareJobs #Talents #WorkingStudents #MunichJobs #BerlinJobs #CareerOpportunities
0
2
4
@msaid_de
MSAID
1 year
๐Ÿ“ฃ Attention students and graduates! Join us at the upcoming job fair #IKOM2024 on June 26th. Discover our open positions and connect with the MSAID team. Don't miss out on these exciting opportunities! #JobFair #SoftwareJobs #MachineLearning #Bioinformatics #Hiring #IKOM
0
3
4
@PastelBio
Pastel BioScience
1 year
Assessment of false discovery rate control in tandem mass spectrometry analysis using entrapment https://t.co/ijMvqCo2Qr --- #proteomics #prot-preprint
0
2
12
@MagnusPalmblad
Magnus Palmblad
1 year
@RalfGabriels presenting ProteomicsML at an @asmsnews workshop. All contributions to this resource are welcome! Check it out on https://t.co/dHy2XxpMZ7
0
8
20
@sgessulat
Siegfried Gessulat
1 year
Interested in how we integrate and harmonize public data to train our ML models at @msaid? Come visit me at W408
0
0
5
@msaid_de
MSAID
1 year
Excited to announce our new all-in-one MSAID SaaS Platform for proteomics. Visit us at booth #200 at #ASMS2024 or sign up at https://t.co/WPt6ytqwpF to try it for free. #TeamMassSpec #Proteomics #MassSpectrometry #ASMS
0
3
7
@MartinFrejno
Martin Frejno
1 year
#TeamMassSpec & #Proteomics, check out our preprint on #CHIMERYS! We are glad we can finally share this with you and look forward to lively discussions on bottom-up proteomics data analysis at #ASMS2024! Join our breakfast seminar to learn more! https://t.co/wyTSiJirsv
Tweet card summary image
biorxiv.org
Proteomic workflows generate vastly complex peptide mixtures that are analyzed by liquid chromatography-tandem mass spectrometry (LC-MS/MS), creating thousands of spectra, most of which are chimeric...
1
8
31