Nikolay Markov @[email protected]

130 posts

Nikolay Markov @nsmarkov@c.im banner
Nikolay Markov @nsmarkov@c.im

Nikolay Markov @[email protected]

@nsmarkov

Doing analysis of (sc)RNA-seq and more in pulmonary diseases at @NM_lung. Infrastructure, visualisation, photography, beginner jazz piano

Chicago เข้าร่วม Ağustos 2012
189 กำลังติดตาม144 ผู้ติดตาม
scverse
scverse@scverse_team·
@nsmarkov @zoe_piran We do, and yes. Whether a specific meeting is recorded/ uploaded is up to the presenter. We're also still a little new at this and don't have a 100% success rate on recording correctly😅 You can find recordings on our youtube channel: @scverse" target="_blank" rel="nofollow noopener">youtube.com/@scverse
English
1
0
1
37
scverse
scverse@scverse_team·
Next Tuesday at 2023-07-11 18:00 CEST will be another community meeting! @zoe_piran will talk about biolord, a deep generative framework for disentangling known and unknown attributes in single-cell data. Looking forward to seeing everyone there!
English
1
3
15
2.2K
Vitalii Kleshchevnikov, PhD
Vitalii Kleshchevnikov, PhD@vitaliikl·
@nsmarkov It has the same issues - attempts to use NLP field methods with minimal changes. Network science perspective is not helpful either due to conflating distinct interaction types.
English
3
0
1
120
Vitalii Kleshchevnikov, PhD
Vitalii Kleshchevnikov, PhD@vitaliikl·
Unpopular opinion: current wave of “language” models is 70-95% hype. Real progress will come from solving real biological & data analysis problems by creating models that represent nuances of both biology & measurement tech. We know about & struggle with lost of such problems.
English
7
2
52
20.8K
Nikolay Markov @nsmarkov@c.im
Nikolay Markov @[email protected]@nsmarkov·
@TheodorisLab This is really great, thank you! 2 questions: I cannot find the code of the model itself on huggingface, could you link it please? And why have you chosen to have cell representation to be average of gene embeddings, rather than adding a separate token, as in BERT? Thank you
English
1
0
1
469
Dmitry Kobak
Dmitry Kobak@hippopedoid·
@lpachter @clured @pavlinpolicar You make some good points, but this particular claim is simply false. There are multiple metrics according to which t-SNE/UMAP perform MUCH better than the elephant. You know that (because I told you), but still choose not to report it. That was my main criticism of your paper.
English
1
0
4
813
Lior Pachter
Lior Pachter@lpachter·
The crazy thing about this tweet is that in his paper on the "art" of dimensionality reduction, @hippopedoid literally recommends starting with two principal components which he claims "injects the global structure into the embedding".
Lior Pachter tweet media
Dmitry Kobak@hippopedoid

@lpachter I'm confused: why did you use `sc .tl.pca(adata, n_comps=2)` in your notebook? Who keeps 2 PCs before doing UMAP? If the point of your exercise is to follow the most standard Scanpy workflow, then keep 50 PCs (n_comps=50), which gives this structureless blob:

English
4
0
15
42.6K
Simona Cristea
Simona Cristea@simocristea·
-- Notations 2 -- N_u - neighborhood nodes: set of neighboring nodes for each node u X_N_u - neighborhood features: set of feature vectors for all neighboring nodes h_u - local function, takes into account the neighborhood h_u = \theta(x_u, X_N_u)
English
2
1
3
778
Simona Cristea
Simona Cristea@simocristea·
Graph Neural Networks (#GNNs) & their applications to life sciences are an exciting #DeepLearning area to discover! But, to develop or apply GNN methods, we first need to understand the maths behind. So, back to basics! Here's a plain language summary of what's behind GNNs👇
Simona Cristea tweet media
English
3
46
231
30.6K
Sasha Berliner
Sasha Berliner@sasha_berliner·
Out everywhere now 💜
Sasha Berliner tweet media
Stanford, CA 🇺🇸 English
2
3
31
0
Nikolay Markov @nsmarkov@c.im
Nikolay Markov @[email protected]@nsmarkov·
I share same concerns: “truth” won't be enough to make people not act upon lies/misinformation (also because there are no facts). Good to see someone proposing other approaches
Carl Miller@carljackmiller

I've heard a lot of people in events and things over the past few weeks calling for a 'western information warfare'. Whilst I understand the sentiment (and the frustration) there are two big, linked reasons why it wouldn't ever work. 1. You don't fight the kind of illicit...

English
0
0
2
0
Malte Luecken
Malte Luecken@MDLuecken·
@rmassonix Can I request a python equivalent on this thread as well, please?
English
1
0
8
0
Ramon Massoni Badosa
Ramon Massoni Badosa@rmassonix·
Hey friends, can someone please develop an R package / function that plots the UMAP1/UMAP2 axis as shown in this plot? I think 90% of us like to use this format in papers but I'm tired to do it with Inkscape. Thanks!!
Ramon Massoni Badosa tweet media
English
19
39
246
0
XD 💙💛
XD 💙💛@XCoustique·
@pepel_klaasa @sirkatrin Я его всегда идентифицировала как запах жжёной подошвы кед ахах
Русский
2
0
0
0
Pepel Klaasa
Pepel Klaasa@pepel_klaasa·
Катя @sirkatrin пьёт какой-то проклятый чай. Он называется лапсанг и на запах он как нефть. Катя пьёт его ещё и с молоком. Никогда не думала, что мне придётся понюхать нефть с молоком, А ВОТ.
Русский
26
1
168
0
XD 💙💛
XD 💙💛@XCoustique·
Я когда-то в подарок партнёру сделала шуточную блок-схему как отличать большое количество европейских языков (было что-то вроде 45-50) чисто по графике и к каким семьям они относятся. Может, послать её этим дятлам
Русский
2
0
18
0
XD 💙💛
XD 💙💛@XCoustique·
*орёт на лингвистическом* Польский, польский не признали, я понимаю там чешский и словацкий не отличили по графике, но зачем-то же были все эти мемы про чтение освежителя воздуха, ну, можно было выучить графику хотя бы Но нет, фоменковщина
Visegrád 24@visegrad24

These Russian soldiers conjecture that the Americans have invented a language in which Ukrainian words are written in the Latin alphabet. The soldiers are clearly clueless that they are holding a Polish RPG-76 Komar (Mosquito). Absolutely unbelievable.

Русский
2
1
20
0
Nikolay Markov @[email protected] รีทวีตแล้ว
XD 💙💛
XD 💙💛@XCoustique·
Я лингвистка и мне надоели мифы,которые используются в пропаганде. Тред с аргументами,вдруг пригодится Структура 1. «рус,укр и бел – диалекты одного языка»(нет) 2. «сограждане на основании языка» (нет) 3. «защита рус языка»(нет) 4. «Зеленский притворяется,что забыл рус»(нет)
Русский
120
2.7K
12.2K
0
Nikolay Markov @nsmarkov@c.im
Nikolay Markov @[email protected]@nsmarkov·
Key takeaway for me: spike of new accounts and account activity since Russia's invasion in Ukraine, apparently correlated with the messages (hard to actually see in the thread). Why important? “The world” is not just US&Europe&China&Japan. Talking with all others is important.
Carl Miller@carljackmiller

When we say Kyiv is winning the information war, far too often we only mean information spaces we inhabit. Pulling apart the most obvious RU info op to date (as we did using semantic modelling), very clear it is targeting BRICS, Africa, Asia. Not the West really at all.

English
0
0
0
0
Nikolay Markov @nsmarkov@c.im
Nikolay Markov @[email protected]@nsmarkov·
@carljackmiller Don't you (grimly) think that this is the future, and the current reframing of information as bullets will only be picked up? So many actors, ceasefire seems impossible, control is bad idea and very expensive. Price of actual verifiable information should increase a lot
English
0
0
0
0
Carl Miller
Carl Miller@carljackmiller·
We are increasingly using militarised terminologies to discuss what happens in information 'spaces'. This might sound hopelessly utopian, but has anyone began to look at what (in very general terms) a ceasefire or disarmament might look like?
Carl Miller@carljackmiller

If information really is a theatre of war (as a writer I find that idea entirely distressing) then attention is probably the closest thing there is to 'territory' that can be won, lost, swapped, contested and basically occupied.

English
5
1
14
0
Nikolay Markov @nsmarkov@c.im
Nikolay Markov @[email protected]@nsmarkov·
@vallens Thank you, this is great. A question: both cells a and b in your example have zero counts of CD24, but scVI models says that it is differentially expressed between them. Where does this information come from?
English
1
0
0
0
Nikolay Markov @[email protected] รีทวีตแล้ว
Alina Leonova
Alina Leonova@AlinaLeonovaSF·
I'm Belarusian. Seeing more and more hate against Belarusians as a result of Lukashenko's participation in Putin's atrocious war against Ukraine, I have a few things to say.
English
1.7K
12.7K
39K
0

ค้นพบ

@NM_Lung @NUFeinbergMed @Ziyou2015 @scverse_team @zoe_piran @vitaliikl @TheodorisLab @hippopedoid