Computational Biologist, PhD student at the School of Computer Science @CarnegieMellon
Pittsburgh, PA
Joined May 2022
- Tweets19
- Following103
- Followers95
- Likes112
Gün Kaynar retweeted
SCS researchers used MHS to run experiments about 3x faster than before, with an AI agent orchestrating a liquid handler, a plate reader, a robotic arm and monitoring cameras across three computers with fundamentally incompatible interfaces.
Today, we're kicking off the first phase of the research preview for Model Hardware Standard (MHS): a new standard for AI agents to safely operate physical equipment in scientific research and advanced manufacturing.
Read more: anthropic.com/news/model-har…
MHS research preview is out. Our team at CMU has been part of it.
We had an AI agent run serial dilution dose-response experiments across a liquid handler, plate reader, and robotic arm with three computers, three incompatible interfaces.
Today, we're kicking off the first phase of the research preview for Model Hardware Standard (MHS): a new standard for AI agents to safely operate physical equipment in scientific research and advanced manufacturing.
Read more: anthropic.com/news/model-har…
Opus figured out each instrument on its own, including one with no API at all (it used the GUI like a human would). Integration took eight hours instead of weeks.
The agent rejected the first run when the curve saturated, adjusted the concentration range, and got a clean fit on the second try.
I wrote up the full story here: gunkaynar.com/blog/mhs-lab-e…
Gün Kaynar retweeted
LYCEUM, our ancient DNA (aDNA) CNV caller is out! aDNA is noisy and high coverage data is scarce, forbidding the use of ML models. The idea is transfer learning from a model that works on another noisy domain - WES, and then fine tuning with aDNA data. biorxiv.org/content/10.1101/…
Gün Kaynar retweeted
Latest MS graduate from the lab is @gunkaynar! He coauthored 3 papers @OUPBioinfo and @NatureComms during his time at CICEKLAB and is now moving on to @CarnegieMellon to seek a PHD degree.
Gün Kaynar retweeted
ECOLE is now featured as an Editors' Highlight pick @NatureComms. The list includes 50 best papers recently published in the journal in the biotechnology and methods area.
nature.com/collections/idhhg…
@BerkveKreb @furkozden @gunkaynar @m_alperyilmaz @calkan_cs @sgqcheng
Happy to share that our new method, ECOLE, for detecting copy number variants on whole exome sequencing data is now published in @NatureComms.
Thanks to CICEKLAB members @BerkveKreb,
@furkozden, @gunkaynar, @m_alperyilmaz and to @calkan_cs.
nature.com/articles/s41467-0…
Gün Kaynar retweeted
@crashius @calkan_cs @BerkveKreb @furkozden @gunkaynar @m_alperyilmaz present ECOLE, a deep learning-based somatic and germline #CNV caller for #WES data, utilising a variant of the transformer architecture. @NatureComms
#BiotechNatureComms
nature.com/articles/s41467-0…
Gün Kaynar retweeted
Happy to share that our new method, ECOLE, for detecting copy number variants on whole exome sequencing data is now published in @NatureComms.
Thanks to CICEKLAB members @BerkveKreb,
@furkozden, @gunkaynar, @m_alperyilmaz and to @calkan_cs.
nature.com/articles/s41467-0…
Gün Kaynar retweeted
Övünmek gibi olmasın, kendisi ablam olur. @aihsanvarol
youtube.com/watch?v=7SeoQrGl…
Gün Kaynar retweeted
At 4pm, @gunkaynar will present our work w/ @mnmsPlatform on analyzing tumor metabolome using biologically-informed neural networks to give online feedback to the surgeon about the predicted survival of the patient @recombccb2023. Preprint: biorxiv.org/content/10.1101/…
Gün Kaynar retweeted
Boğaziçi Üniversitesi'nde mezuniyet törenleri dün sona erdi. Törenler ile ilgili yaşanan ve kayyum yönetimin başrolü oynadığı tüm rezillikleri derli toplu durması açısından bu dizide anlatacağım. (1/20)
Gün Kaynar retweeted
Our manuscript on brain tumor margin detection during surgery is published in @OUPBioinfo. We show that incorporating prior knowledge on metabolite peak region s improves tumor detection using HRMAS NMR signal. bit.ly/391vlzA Another nice collab. with @mnmsPlatform.