Nikolay Oskolkov
oskolkov.bsky.social
Nikolay Oskolkov
@oskolkov.bsky.social
Group Leader (PI) at LIOS, Riga, Latvia

PhD in theoretical physics. Research interests: computational biology, statistics, machine learning, aDNA / eDNA, ancient metagenomics.

Homepage: https://nikolay-oskolkov.com/
The machine learning multiOmics integration methodology being developed in my research group at LIOS targetwise.osi.lv/research/met... has recently been presented at the Precision Medicine PMNET forum pmnetforum.com/speakers/
October 22, 2025 at 7:47 AM
It has been a wonderful month in Riga, very excited building the Metabolic Research Group at LIOS and interacting with my new fantastic colleagues. Thanks everyone for your hospitality!
October 17, 2025 at 8:32 AM
The Metabolic Research Group @TARGETWISE_ is growing, and we are very happy to welcome Daniel Rivas, MD and PhD in AI, the first postdoctoral fellow in the group, who is an amazing Machine Learning expert! So great to have you on board, Daniel!
October 15, 2025 at 4:31 PM
Our method, GENome EXogenous (GENEX) sequence detection, for identifying microbial-like regions in eukaryotic reference genomes is online in GigaScience track.smtpsendmail.com/9032119/c?p=...
October 4, 2025 at 9:20 AM
Nature discusses our article on mammoth microbiome www.nature.com/articles/d41...
September 4, 2025 at 5:12 AM
Our paper on ancient mammoth microbiome analysis is out in Cell. Congratulations Ben Guinet, Tom van der Valk and all the co-authors! authors.elsevier.com/c/1liYML7PXu...
September 3, 2025 at 3:49 AM
Our paper on ancient infectious diseases in Eurasia is out in Nature. Congratulations Martin Sikora, Eske Willerslev and all the co-authors!
July 10, 2025 at 4:49 AM
Teaching stats and machine learning at #RaukR is always fun, thanks everyone for the great discussion!
June 19, 2025 at 1:30 PM
Glory Victory Day!
May 9, 2025 at 8:45 AM
If you use Kraken for profiling microbial / environmental / ancient metagenomic data, please check this preprint where I benchmark filtering strategies of Kraken outputs on simulated microbial and environmental metagenomic samples, and provide recommendations for optimal filtering thresholds
April 10, 2025 at 6:57 AM
Attention ancient environmental DNA folks. In our recent preprint, we provide coordinates of microbial-like regions (contaminated or evolutionary conserved or convergent) for 4,294 eukaryotic reference genomes.
March 24, 2025 at 10:33 AM
For comparison, here is your code from the blogpost run 9 times on different seeds from 1 to 9. Even here the "artificial clustering" in not always clear
February 13, 2025 at 11:30 AM
@rafalab.bsky.social also, did you try increasing the number of data points? This is what I get running your code on 10 000 data points with seeds from 1 to 9. The artificial separation is not visible. So isn't the artificial separation you show in your blogpost just the pitfall of small numbers?
February 13, 2025 at 11:27 AM
For comparison, if you start with 3-dimensional data as you state in your post ("three correlated random variables") and use "u <- umap(x)" instead, this is the output (with "dist" on the left, without "dist" on the right)
February 13, 2025 at 9:23 AM
I am giving a seminar discussing recent criticism of using UMAP in All of Us paper in Nature doi.org/10.1038/s415.... I will compare peculiarities of UMAP for single cell biology vs. population genomics applications. Feel free to join if you are interested, www.scilifelab.se/event/is-uma....
February 13, 2025 at 8:19 AM
aMeta computational method for ancient metagenomic analysis genomebiology.biomedcentral.com/articles/10.... was presented at Evolution in Sweden 2025 conference. Thanks a lot everyone for you feedback!
January 18, 2025 at 7:10 AM
Our paper on ancient Pseudogymnoascus fungus found in the gut of Tyrolean Iceman (Ötzi) is published in BMC Genomics. Congratulations Luigi Borruso, Anna Sandionigi and co-authors!
December 20, 2024 at 9:09 AM