Vadim Demichev

1.5K posts

Vadim Demichev

Vadim Demichev

@DemichevLab

Group Leader at @ChariteBerlin. Proteomics technologies and applications, DIA-NN author.

Berlin, Germany Katılım Kasım 2018
2.4K Takip Edilen3.7K Takipçiler
Sabitlenmiş Tweet
Vadim Demichev
Vadim Demichev@DemichevLab·
We are delighted to present DIA-NN 2.5. Redefining LC-MS proteomics with up to 70% more proteins identified and powerful capabilities for statistics, visualisation and biological interpretation.
Vadim Demichev tweet media
English
2
9
46
2.7K
Vadim Demichev
Vadim Demichev@DemichevLab·
Great to see DIA-NN 2.5 Enterprise already used in cutting edge publications. TomAP-MS: an improved tomato lectin affinity purification-based mass spectrometry workflow enabling ultra-deep plasma proteomics: over 7500 proteins from plasma on Orbitrap Astral at 16 SPD and 6500 at 50 SPD.
Vadim Demichev tweet media
English
1
1
20
1.1K
Vadim Demichev
Vadim Demichev@DemichevLab·
We have released DIA-NN 2.5.1 update with minor GUI changes. We would be grateful for feedback and suggestions on the new statistics, visualisation and biological interpretation functionality introduced in DIA-NN 2.5. What can we make better? Any extras we can add?
Vadim Demichev tweet media
English
3
1
14
567
Vadim Demichev
Vadim Demichev@DemichevLab·
Release notes: github.com/vdemichev/DiaN… The new DIA-NN 2.5 is available for download from the same secure link (Enterprise customers) or GitHub (Academia version for non-profits).
English
0
1
6
312
Vadim Demichev
Vadim Demichev@DemichevLab·
We are delighted to present DIA-NN 2.5. Redefining LC-MS proteomics with up to 70% more proteins identified and powerful capabilities for statistics, visualisation and biological interpretation.
Vadim Demichev tweet media
English
2
9
46
2.7K
Vadim Demichev
Vadim Demichev@DemichevLab·
The best security practice used to be 'keep your dependencies up to date.' I suspect in the future it will be: take an old known-good source snapshot → AI auto-audit + patch → review → never touch it again.
Andrej Karpathy@karpathy

Software horror: litellm PyPI supply chain attack. Simple `pip install litellm` was enough to exfiltrate SSH keys, AWS/GCP/Azure creds, Kubernetes configs, git credentials, env vars (all your API keys), shell history, crypto wallets, SSL private keys, CI/CD secrets, database passwords. LiteLLM itself has 97 million downloads per month which is already terrible, but much worse, the contagion spreads to any project that depends on litellm. For example, if you did `pip install dspy` (which depended on litellm>=1.64.0), you'd also be pwnd. Same for any other large project that depended on litellm. Afaict the poisoned version was up for only less than ~1 hour. The attack had a bug which led to its discovery - Callum McMahon was using an MCP plugin inside Cursor that pulled in litellm as a transitive dependency. When litellm 1.82.8 installed, their machine ran out of RAM and crashed. So if the attacker didn't vibe code this attack it could have been undetected for many days or weeks. Supply chain attacks like this are basically the scariest thing imaginable in modern software. Every time you install any depedency you could be pulling in a poisoned package anywhere deep inside its entire depedency tree. This is especially risky with large projects that might have lots and lots of dependencies. The credentials that do get stolen in each attack can then be used to take over more accounts and compromise more packages. Classical software engineering would have you believe that dependencies are good (we're building pyramids from bricks), but imo this has to be re-evaluated, and it's why I've been so growingly averse to them, preferring to use LLMs to "yoink" functionality when it's simple enough and possible.

English
1
0
1
411
Vadim Demichev retweetledi
Tenzer Lab
Tenzer Lab@TenzerLab·
Our benchmarks reveal inflated true FDR (3–5%) when modified peptidoforms are considered and highlight peak-matching errors in match-between-runs analyses. Great collaboration with the labs of Andreas Hildebrandt and Mathias Wilhelm. (2/2) Preprint: researchsquare.com/article/rs-903…
English
1
2
6
1.1K
Vadim Demichev
Vadim Demichev@DemichevLab·
What an amazing work! Regular entrapment FDR validation can be circumvented in a number of ways. But here comes simulated data with known ground truth. Great to see DIA-NN 2.0 excel at sensitivity while controlling FDR and false localisation rates.
Vadim Demichev tweet media
Tenzer Lab@TenzerLab

Our benchmarks reveal inflated true FDR (3–5%) when modified peptidoforms are considered and highlight peak-matching errors in match-between-runs analyses. Great collaboration with the labs of Andreas Hildebrandt and Mathias Wilhelm. (2/2) Preprint: researchsquare.com/article/rs-903…

English
0
0
8
561
Tenzer Lab
Tenzer Lab@TenzerLab·
New preprint out! Timsim: a framework to simulate native #timsTOF dda-PASEF and dia-PASEF datasets with full ground-truth annotation, enabling precise benchmarking of #FDR and phosphosite localization errors in #proteomics and #immunopeptidomics workflows. (1/2)
English
2
0
4
229
Vadim Demichev retweetledi
Matthias Mann Lab
Matthias Mann Lab@labs_mann·
Now in @CellCellPress: We profiled cerebrospinal fluid proteomes from 5,000 neurology patients by mass spectrometry — mapping protein changes across stroke, brain cancer, infections & autoimmune diseases, revealing shared and disease-specific signatures. sciencedirect.com/science/articl…
Matthias Mann Lab tweet media
English
2
38
140
9.6K
Vadim Demichev
Vadim Demichev@DemichevLab·
Solving the computational challenge of phosphoproteomics with 𝐏𝐡𝐨-𝐓𝐢𝐩: One-Pot Dephosphorylation for Rapid and Sensitive Analysis of DIA Phosphoproteomics Data. Now out in Analytical Chemistry! Makes predicted phosphopeptide libraries 10x-20x smaller. Link below.
Vadim Demichev tweet media
English
1
2
12
501
Vadim Demichev retweetledi
Matthias Mann Lab
Matthias Mann Lab@labs_mann·
Ever wondered how the human liver looks like at single-cell, spatial protein resolution? We used single-cell Deep Visual Proteomics to map human liver zonation at the protein level - one hepatocyte at a time. Our paper is out in @NatMetabolism! nature.com/articles/s4225…
English
2
17
76
6.6K
Vadim Demichev retweetledi
Matthias Mann Lab
Matthias Mann Lab@labs_mann·
Discovery proteomics → clinical diagnostics? ADAPT-MS makes it possible by adapting to each sample's protein coverage on-the-fly. No imputation, no fixed panels, multiple diagnostic questions from one measurement. Out now in @NatureComms. nature.com/articles/s4146…
English
0
6
36
2.9K
Vadim Demichev
Vadim Demichev@DemichevLab·
@OdedRechavi @X @grok Indeed. I have switched to LinkedIn almost exclusively for getting news about papers. X is full of 'funny stuff' and 'political stuff', I tried to shift the focus of what I see to AI and coding by selecting the relevant topics, but still get bombarded by irrelevant things.
English
0
0
5
188
Oded Rechavi
Oded Rechavi@OdedRechavi·
Two options: 1. The @X algorithm is just broken 2. The algorithm intentionally ignores scientists (news, tech & entertainment less effected) It’s probably #1. We heard that they were going to fix it (@grok), but this didn’t happen. Exposure to science is down >90%.
English
31
31
303
28.5K