Xin Lai (@laixinking) header
Xin Lai (@laixinking) profile picture

A Systems Biologist with passion for research, photograph, and cook.

Finland
Joined December 2009
New research in @CMLS_Journal provides the first systematic look at SOX10-mediated miRNA regulation in Uveal Melanoma (UM). 🔗 Read the full study here: rdcu.be/fjbeA #UvealMelanoma #CancerResearch #SystemsBiology #miRNA #Bioinformatics #Oncology #PrecisionMedicine
1
1
56
We develop #RingNet to visualize #multi-modal data in #networks. It is powerful to visualize your #cohort and #single-cell #omics data. All you need are the csv files for your data and networks, and RingNet can support svg, pdf output. Try it out at fip-128-214-252-149.kaj.pout…
1
1
1
78
The latest research from Guo's lab at @theNCI shows that one hotspot mutation in #DICER1's RNase IIIb can lead to loss and gain of function 5p-miRNA and the gain of function of 3p-miRNA. rdcu.be/eUxiq
57
Seedless binding in #miRNAs can also lead to gene repression as shown with miR-155 and miR-124, though this unconventional binding accounts ~1% of miRNA binding. See Bartel lab's new paper for details. Functional microRNA targeting without seed pairing academic.oup.com/nar/article…
38
New preprint on IDH-mutant #astrocytomas: We analyzed rare patient samples showing grade 2-3→4 progression post-therapy. Key findings: #ncRNAs drive astrocyte dedifferentiation; tumor suppressor #miRNA let-7b-3p shows recurrent deletions; #lncRNA NEAT1 regulates ECM organization
71
We built HallmarkGraph to be transparent. By integrating gene-level features with cancer hallmarks, we can now provide clear explanations for our model's predictions. #InterpretableAI #ComputationalOncology #MedicalAI
1
31
HallmarkGraph achieves up to 99% accuracy in classifying tumor subtypes. Its high performance and interpretability can help us understand and target complex cancers more effectively. #DeepLearning #ComputationalBiology #AIforGood #HallmarksOfCancer
1
19
We validated HallmarkGraph on over 10,000 tumor samples across 26 cancers and 405 subtypes up to eight levels. Our model is effective and scalable, paving the way for more robust cancer classification. #DataScience #GraphNeuralNetworks #Oncology #Genomics
1
2
31
Introducing HallmarkGraph, the first biologically-informed graph neural network designed for hierarchical tumor-subtype classification. We’re integrating biological principles to advance precision oncology. #AI #MolecularTumorSubtype #TumorClassification #NetworkMedicine
1
19
The forum's emphasis on translating academic research into practical applications through industry partnerships reflects the urgent need for scalable, data-driven healthcare solutions that can be deployed globally to improve patient outcomes and population health management.
24