Egor Marin
@marinegor.bsky.social
ML Scientist @ ENPICOM B.V. (Den Bosch, Netherlands)
computational biology, ML, protein design, cheminformatics, fancy dev tooling, tinge of bouldering
https://marinegor.dev
computational biology, ML, protein design, cheminformatics, fancy dev tooling, tinge of bouldering
https://marinegor.dev
that was actually my motivation behind choosing a uni: I deliberately went for the one offering most flexibility throughout the bachelor, and ended up using it to its maximu.
November 23, 2025 at 6:40 PM
that was actually my motivation behind choosing a uni: I deliberately went for the one offering most flexibility throughout the bachelor, and ended up using it to its maximu.
damn that's complicated! I guess i'm too lazy to go that deep😁
ah, and caps is language switching for me, also insanely useful.
ah, and caps is language switching for me, also insanely useful.
November 23, 2025 at 10:49 AM
damn that's complicated! I guess i'm too lazy to go that deep😁
ah, and caps is language switching for me, also insanely useful.
ah, and caps is language switching for me, also insanely useful.
hey Pablo, congrats on the publication!
I imagine you started working on it before MDAnalysis introduced parallelization, so I wonder if you'd be interested in implementing it for eRMSF as described here: docs.mdanalysis.org/stable/docum...
I imagine you started working on it before MDAnalysis introduced parallelization, so I wonder if you'd be interested in implementing it for eRMSF as described here: docs.mdanalysis.org/stable/docum...
4.2.4. Parallel analysis — MDAnalysis 2.10.0 documentation
docs.mdanalysis.org
November 22, 2025 at 11:28 PM
hey Pablo, congrats on the publication!
I imagine you started working on it before MDAnalysis introduced parallelization, so I wonder if you'd be interested in implementing it for eRMSF as described here: docs.mdanalysis.org/stable/docum...
I imagine you started working on it before MDAnalysis introduced parallelization, so I wonder if you'd be interested in implementing it for eRMSF as described here: docs.mdanalysis.org/stable/docum...
<<<< binding right cmd to escape
November 22, 2025 at 11:02 PM
<<<< binding right cmd to escape
please tell me there was "rust is blazingly fast" option
November 17, 2025 at 8:45 PM
please tell me there was "rust is blazingly fast" option
I checked Plank and Einstein obituaries -- they've referred as Dr all throughout. But I guess it's, unfortunately, not the worst disgrace she's had throughout.
November 9, 2025 at 4:49 PM
I checked Plank and Einstein obituaries -- they've referred as Dr all throughout. But I guess it's, unfortunately, not the worst disgrace she's had throughout.
within 24h, someone who I never comes in with a chunky PR that a) is more concise than my code, b) fixes all of my issues, c) also fixes relevant documentation for that. All I had left to do is to add this person to the changelog.
Feels unexpectedly good, I should say.
Feels unexpectedly good, I should say.
November 6, 2025 at 12:29 PM
within 24h, someone who I never comes in with a chunky PR that a) is more concise than my code, b) fixes all of my issues, c) also fixes relevant documentation for that. All I had left to do is to add this person to the changelog.
Feels unexpectedly good, I should say.
Feels unexpectedly good, I should say.
I immediately wonder if there's any preference for antibody fold around heavy/light chain variable fragment length🤔
October 27, 2025 at 9:19 PM
I immediately wonder if there's any preference for antibody fold around heavy/light chain variable fragment length🤔
>Send me a message for any questions
it seems that your DMs are closed, could you double-check your privacy settings please?
it seems that your DMs are closed, could you double-check your privacy settings please?
October 10, 2025 at 1:49 PM
>Send me a message for any questions
it seems that your DMs are closed, could you double-check your privacy settings please?
it seems that your DMs are closed, could you double-check your privacy settings please?
yep, it was Andre Geim, IgNobel for diamagnetic levitation (funny flying frog, it's also on youtube) and Nobel for graphene
October 3, 2025 at 10:41 PM
yep, it was Andre Geim, IgNobel for diamagnetic levitation (funny flying frog, it's also on youtube) and Nobel for graphene
people at apple rn
September 25, 2025 at 8:01 AM
people at apple rn
impressive maps!
aren't you worried that someone might guess your target based on these maps and e.g. scoop you?🫣
aren't you worried that someone might guess your target based on these maps and e.g. scoop you?🫣
September 22, 2025 at 9:07 AM
impressive maps!
aren't you worried that someone might guess your target based on these maps and e.g. scoop you?🫣
aren't you worried that someone might guess your target based on these maps and e.g. scoop you?🫣
and afaik it's implementation-dependent, so for example pypy could've done things differently
September 21, 2025 at 2:59 PM
and afaik it's implementation-dependent, so for example pypy could've done things differently
it honestly looks like the same shape in different colors, copypasted across the plot😁
September 19, 2025 at 10:35 AM
it honestly looks like the same shape in different colors, copypasted across the plot😁
that's a t-shirt idea right here
September 18, 2025 at 1:27 PM
that's a t-shirt idea right here
oh that's a great one
there was also a weak negative correlation between model quality and a journal impact factor, I recall -- high(est) impact papers tend to focus on more challenging targets, which results in lower resolution and worse QC scores for the model.
there was also a weak negative correlation between model quality and a journal impact factor, I recall -- high(est) impact papers tend to focus on more challenging targets, which results in lower resolution and worse QC scores for the model.
September 14, 2025 at 10:29 AM
oh that's a great one
there was also a weak negative correlation between model quality and a journal impact factor, I recall -- high(est) impact papers tend to focus on more challenging targets, which results in lower resolution and worse QC scores for the model.
there was also a weak negative correlation between model quality and a journal impact factor, I recall -- high(est) impact papers tend to focus on more challenging targets, which results in lower resolution and worse QC scores for the model.
I think the first in line was some ide that forked PyCharm community edition, but I forgot the name :)
September 8, 2025 at 8:13 PM
I think the first in line was some ide that forked PyCharm community edition, but I forgot the name :)
on a more serious note, you might try referring to the existence of a strong Rust community in the country of application, and say that it'll help you to soften the bus factor and grow the community further🤷♂️
September 3, 2025 at 11:03 AM
on a more serious note, you might try referring to the existence of a strong Rust community in the country of application, and say that it'll help you to soften the bus factor and grow the community further🤷♂️
I honestly believe that the "career impact" section of a grant application should look like a single sentence saying "no grant -- no career"
September 3, 2025 at 11:00 AM
I honestly believe that the "career impact" section of a grant application should look like a single sentence saying "no grant -- no career"
so you're saying that perhaps this self-censorship is limiting reviewers from writing good reviews?
August 21, 2025 at 8:22 AM
so you're saying that perhaps this self-censorship is limiting reviewers from writing good reviews?
>diffusion-based antibody-specific NNs wouldn't do this?
why though? when it comes to antigens, they definitely memorise the interface, as per recent independent benchmark😁
why though? when it comes to antigens, they definitely memorise the interface, as per recent independent benchmark😁
August 16, 2025 at 5:54 AM
>diffusion-based antibody-specific NNs wouldn't do this?
why though? when it comes to antigens, they definitely memorise the interface, as per recent independent benchmark😁
why though? when it comes to antigens, they definitely memorise the interface, as per recent independent benchmark😁
oh sorry, by models I meant which NNs, not antibody structures
August 15, 2025 at 2:57 PM
oh sorry, by models I meant which NNs, not antibody structures