Bell Eapen retweeted
Elated to share that the Weinstock lab just got its 1st R01!!
We are SO grateful to the hardworking NIH staff for tirelessly pushing to get those last NOAs out
Bell Eapen retweeted
Genomic Tokenizer: Toward a biology-driven tokenization in transformer models for DNA sequences biorxiv.org/content/10.1101/… 🧬🖥️🧪 github.com/dermatologist/gen…
Bell Eapen retweeted
Genomic Tokenizer: Toward a biology-driven tokenization in transformer models for DNA sequences
1. Genomic Tokenizer (GT) introduces a biologically grounded approach to DNA sequence tokenization, aligning with the central dogma of molecular biology by using codons—three-letter nucleotide sequences—as the core unit of tokenization.
2. Unlike traditional character or k-mer tokenizers, GT recognizes start and stop codons, assigns identical tokens to synonymous codons, and treats introns and out-of-frame regions as UNK tokens, reducing vocabulary size while preserving biological relevance.
3. GT is implemented within the HuggingFace tokenizer framework, enabling seamless integration into existing transformer-based DNA analysis pipelines and support for tasks like masked language modeling and sequence classification.
4. The tokenizer supports customizable start/stop codons and intron treatment, making it adaptable for different organisms, including prokaryotes and mitochondrial genomes.
5. In classification experiments using a lung cancer-related variant dataset, GT showed greater robustness to long sequence lengths compared to character tokenization, and outperformed it in longer sequence tasks.
6. While byte-pair encoding (BPE) achieved the highest overall performance, its large vocabulary comes with high computational cost. GT balances biological insight with computational efficiency and a compact vocabulary.
7. GT tokenization avoids issues of redundancy and information leakage in masked language modeling that are common in overlapping k-mer tokenizers, leading to cleaner training signals and potentially better generalization.
8. The biological foundation of GT allows it to better model frame-shift mutations, synonymous substitutions, and stop-gain variations—key features in predicting phenotypic impact from genetic data.
9. Preliminary comparisons highlight GT’s strength in biological modeling and suggest potential advantages for foundational model training across genomics tasks when compared with purely data-driven tokenizers.
10. GT is open-source, installable via PyPI, and encourages broader exploration across genomic datasets and transformer architectures, including long-context models such as HyenaDNA.
💻Code: github.com/dermatologist/gen…
📜Paper: biorxiv.org/content/10.1101/…
#Genomics #Tokenization #Transformers #Bioinformatics #DNASequence #LLM #Codon #MaskedLanguageModel #DeepLearning #HuggingFace
Bell Eapen retweeted
Researchers at @DESE_IISc led by Hardik J Pandya, with collaborators at @aiimsrishi, have developed an #algorithm that can sift through #EEG data & help identify the type and occurrence of #Epilepsy.
iisc.ac.in/events/detecting-…
#IIScresearch
Bell Eapen retweeted
A paper from @beapen @ferozkal and @amanzadermat that will be interesting to people in AI, Machine Learning etc. Its free open access from this link 👇
link.springer.com/article/10…
#dermatology #Artificial_Intelligence #MachineLearning #imaging #skinimaging – at Perinthalmanna, India
Bell Eapen retweeted
PyPI fastapi-featureflags: very simple implementation of feature flags for @FastAPI:
⊛ Minimum configuration required ⚙️
⊛ No unnecessary dependencies 👻
⊛ Does its job ✅
pypi.org/project/fastapi-fea…
Bell Eapen retweeted
Not a good sign in the US when the number of new cases per day is
—3 times the number of new people vaccinated
—Hundreds of thousands more than the number of people getting boosted
[Omicron is not a live attenuated vaccine]
Bell Eapen retweeted
New! How vaccine-induced antibodies to PF4 trigger the thrombotic events seen in #VITT
@NatRevImmunol #BrighterWorld @McMasterU
Bell Eapen retweeted
The last 4 months have been life-changing, thank you to @NatureCareers for featuring my story and thank you to other patient partners for spearheading patient engagement in research.
@rheum_covid #COVID19 #PatientExperience
nature.com/articles/d41586-0…
Bell Eapen retweeted
ML researchers: We have a model you can use.
Clinician: Does it work in this situation?
ML: Hold on… we need to go run some experiments.
…
ML researchers: The experiments were a success!
…
Clinician: Just one question. How many mice?
ML researchers: ?
Clinician: ?
Bell Eapen retweeted
So with 84% 6 month efficacy for Pfizer and 93% for moderna - can we stop talking about boosters before global vaccine campaigns catch up even a little?
Yes there will be breakthroughs (not 100%)
There will also be significant complications of those with 0% protection.
Moderna says its COVID-19 shot remains 93% effective 4-6 months after second dose reut.rs/3fBdOP0
Bell Eapen retweeted
Hot off the press: So excited about this new work extracting knowledge graphs from radiology reports--something I have wanted to pursue for many years. Congrats to a fantastic team of @stanford students working with collaborators from @VinBrainAI.
arxiv.org/abs/2106.14463
Bell Eapen retweeted
Effortless Bayesian Deep Learning.
#Serverless #DataScientist #Linux #Statistics #Programming #Coding #100DaysofCode #BigData #Analytics #DataScience #AI #MachineLearning #IoT #IIoT #Python #RStats #TensorFlow #Java #JavaScript #ReactJS #CloudComputing
Effortless Bayesian Deep Learning. #BigData #Analytics #DataScience #AI #MachineLearning #IoT #IIoT #Python #RStats #TensorFlow #Java #JavaScript #ReactJS #CloudComputing #Serverless #DataScientist #Linux #Statistics #Programming #Coding #100DaysofCode
https://xitter.gq/t.co/e0gZAbSZ9N
Bell Eapen retweeted
People are thinking our potential third wave is going to be driven by the variants - but I think this misses the forest for the trees. To add context - the UK saw population effects from a variant explosion in the UK with a lower level of variants. 1/
Bell Eapen retweeted
Artificial intelligence in dermatology: A practical introduction to a paradigm shift idoj.in/article.asp?issn=222…
via @beapen
#ai #dermatology #dermatologia
Bell Eapen retweeted
Replying to @beapen
I was telling something similar to a friend last week..that the response protocol for Covid-19, could had it been made more scientifically correct in line with Operations Management :)
Bell Eapen retweeted
Presenting research reflexivity in your PhD thesis journals.rcni.com/nurse-rese… @K_Higginbotham1 @profbenlight @beapen
Bell Eapen retweeted
COVID-19 Research in Brief: December, 2019 to June, 2020 nature.com/articles/d41591-0…
Bell Eapen retweeted
Getting things done with #AI - Just published: Yottixel is the first commercial and efficient #image search engine for #histopathology to search for patches and whole-slide images to be offered by @hurondp.
URL: lnkd.in/eQi_9jU




































