Glad to see our work on multi-modal #cfDNA sequencing published in Cell Reports Medicine. We analyzed genomic and fragmentomic data from sWGS of >600 cancer patients, including early-stage lung and esophageal carcinoma. #liquidbiopsy doi.org/10.1016/j.xcrm…

Happy to share! Our latest publication in Genome Biology unveils how BORIS/CTCFL epigenetically reprograms transcriptionally inactive intronic CTCF sites into alternative promoters: genomebiology.biomedcentral.com/articles/10.11…

A short tutorial for all those who are writing their coursework/articles using the Overleaf (LaTeX). Seems easier to reference too compared to EndNote youtube.com/watch?v=Ubi1lg…

I spent almost a week training a Logistic Regression on EEG data manually from scratch, 2 days to feed the data to a CNN model and 2 days to write a report pulling all-nighters in Ramadan. Having no computing background hurts...

The Neural Networks coursework submitted, yeeey! Now I have the last assignment with at least 4 ML models to train/predict the depression related outcome and an equally daunting report ( I dislike writing with so much passion :)). Hope everyone is enjoying the Easter Break🪺.

Our paper on nucleosome repositioning in breast cancer is now published in Clinical Epigenetics! …lepigeneticsjournal.biomedcentral.com/articles/10.11… ▶️Nucleosome gain marks deregulated TFs ▶️Shorter distances between nucleosomes in cancer ▶️NRL affected by gene expression, methylation, H1 histones, repeats

Text analytics assignment was released and the feeling I got " the lecturer threw me into the Niagara Falls for a free fall" or " am I free floating in an alternative universe". I think my PhD was way easier despite many hours in the lab 7/7 days.

Our paper with Helle Jørgensen's lab Helle Jørgensen Lab on gene regulatory networks controlling the proliferation of vascular smooth muscle cells is out. Congratulations to Helle, Jordi, our joint postdoc Sebnem and the rest of the team!

I hope this is the review that finally and definitively clarifies what this mark does 🙏 | ⁦Helin Lab⁩ Roles of H3K4 methylation in biology and disease sciencedirect.com/science/articl…

People are enjoying their ice-creams and I am staring with jealousy while scratching my head and asking why did I choose an optional course in computational finance and stuck with this Markowitz portfolio optimization for investment?!!

Working on numerical data was hard, I thought. Wait till you start working with textual and fine-tuning small models like DistilBert and get stuck for hours tweaking and debugging. On a hunt for models to analyze RNA-seq data using NN. No holidays ahead....

For the past 3days I have been training dis.SVM and unsup. models and felt proud of myself for achieving good metrics. Then I looked up examples , where models disagreed in predicting correctly. Discoverd pre-processing step "butchered" the data, yet the metrics are so 🔥.

Fellow scientists, how can you do all the bioinformatics work without relying on a cluster (if the software package is not there) ? There must be an easier way of building an index, aligning and mapping files in Jupiter notebooks or somewhere else?

⚠️ CTCF binding landscape is established by the epigenetic status of the nucleosome well-positioned relative to CTCF motif orientation BioRxiv Sep 26, Md Tajmul et al with Vladimir B. Teif, Victor Lobanenkov & Elena Pugacheva doi.org/10.1101/2024.0… biorxiv.org/content/10.110…

Please RT and spread the word! Singh lab is expanding and still time to apply online. 2xPDRAs vacancies.essex.ac.uk/tlive_webrecru… vacancies.essex.ac.uk/tlive_webrecru… 1 Research Assistant vacancies.essex.ac.uk/tlive_webrecru… Please feel free to email me ([email protected]) for informal inquiries.

I am so used to installing a software thru running an .exe file. Some require Linux env to install and Windows 10/11 allows to run Linux environment via WSL

♨️ A unified model of gene expression control by cohesin and CTCF ☞ bioRxiv doi.org/10.1101/2024.1… version posted December 3rd 2024

Alibaba’s New AI Beat DeepSeek! Alibaba just launched Qwen2.5, its most powerful AI model yet, which outperforms DeepSeek, OpenAI, and Google. Here’s why it’s a game-changer and why the US government should take notice: 🧵