Avner Schlessinger

While AI models like AlphaFold can predict biomolecular structures, drug failures often occur due to incorrect targets or patient selection, not structural issues. San Francisco-based NOETIK has released an insightful technical report on the Oncology Counterfactual Therapeutics Oracle (OCTO), an AI model that uses a transformer-based architecture trained on multimodal data from patient tumors, including protein staining, gene expression, DNA sequencing, and structural markers. Its innovative design integrates diverse biological data to predict patient-specific treatment responses for complex diseases like cancer. With features like structured multimodal masking and visual prompting, OCTO can infer intricate biological relationships and simulate therapeutic interventions, aiming to enhance precision medicine and streamline drug development. Read a detailed report about the new Foundation model via the link in the comments. Also, subscribe to Where Tech Meets Bio (Substack Newsletter) to not miss our weekly "tech+bio" roundups sent out on Sundays. #foundationmodel #drugdiscovery #biotech Image credit: NOETIK

172

15 Comments

This excellent review by Choi et. al. describes recent technological and intellectual property trends in antibody-drug conjugate research. Quoting from the abstract: "Antibody–drug conjugate (ADC) therapy, an advanced therapeutic technology comprising antibodies, chemical linkers, and cytotoxic payloads, addresses the limitations of traditional chemotherapy. This study explores key elements of ADC therapy, focusing on antibody development, linker design, and cytotoxic payload delivery. The global rise in cancer incidence has driven increased investment in anticancer agents, resulting in significant growth in the ADC therapy market. Over the past two decades, notable progress has been made, with approvals for 14 ADC treatments targeting various cancers by 2022. Diverse ADC therapies for hematologic malignancies and solid tumors have emerged, with numerous candidates currently undergoing clinical trials. Recent years have seen a noteworthy increase in ADC therapy clinical trials, marked by the initiation of numerous new therapies in 2022. Research and development, coupled with patent applications, have intensified, notably from major companies like Pfizer Inc. (New York, NY, USA), AbbVie Pharmaceuticals Inc. (USA), Regeneron Pharmaceuticals Inc. (Tarrytown, NY, USA), and Seagen Inc. (Bothell, WA, USA). While ADC therapy holds great promise in anticancer treatment, challenges persist, including premature payload release and immune-related side effects. Ongoing research and innovation are crucial for advancing ADC therapy. Future developments may include novel conjugation methods, stable linker designs, efficient payload delivery technologies, and integration with nanotechnology, driving the evolution of ADC therapy in anticancer treatment."

171

5 Comments

https://lnkd.in/gaX7bWa5 Happy to publish the article titled "Non-Nucleoside Lycorine-Based Analogs as Potential DENV/ZIKV NS5 Dual Inhibitors" discusses the use of lycorine analogs to target the NS5 RNA-dependent RNA polymerase of dengue and Zika viruses. Through virtual screening, molecular dynamics, and ADMET analysis, the study identified 12 promising lycorine analogs, which showed superior binding affinities compared to ribavirin. These analogs, especially those with distinct enantiomers, present promising avenues for antiviral development against both viruses. Great work with Luke Achenie in Virginia Tech, Paola Andrea Caicedo Burbano in ICESI, Andrés Felipe Vásquez Jiménez, Maria Francisca Villegas-Torres, Yasser Hayek Orduz, Adrian_Camilo Rodríguez_Ararat, Felipe Sierra Hurtado

26

🚀 Interesting work on applying LLMs in drug discovery where the authors used an LLM to generate indications from SMILES and generate SMILES given an. indication! While more work is needed ... it will be very powerful if we could generate SMILES based on an a disease description only! Follow for more daily news

5

Google DeepMind's AlphaFold 3 has expanded predictive capabilities from proteins to a broad spectrum of biomolecular interactions, including DNA, RNA, ligands, and more, contributing to key applications, such as drug discovery. However, the publication of AlphaFold 3 without open source code has sparked a larger conversation within the scientific community around the changing landscape of research tools and the role of journals in communicating science to ensure reproducibility and accessibility to allow further scientific progress. Many thanks to Roland Dunbrack, Stephanie Wankowicz, PhD, and Mohammed AlQuraishi for speaking with me. Read my full article in our peer review journal, GEN Biotechnology: https://lnkd.in/gj2ED2kv ..and on GEN Edge: https://lnkd.in/g9gtUsUT #AI #AlphaFold #Biotechnology #DrugDiscovery

29

1 Comment

Excited to share our team's latest 2024 publication in Cancers! 🧬🔬 3️⃣ Hot off the press: "Gedatolisib, a Pan-PI3K/mTOR Inhibitor, Shows Superior Efficacy to Single-Node PAM Inhibitors in Gynecologic Cancer Models" just published in Cancers! https://lnkd.in/gHdjCa9D A highlight of this project for me - collaborating with Arul Menon, a brilliant intern from UC Berkeley. It was incredibly fun and rewarding to navigate the maze together - always amazes me how skills evolve and intersect in unexpected ways, and how collaboration sparks innovation! 🧮 The Science Behind our analysis: We utilized the Slingshot algorithm, which employs a Minimum Spanning Tree (MST) approach and Mahalanobis distance to infer cancer progression from static transcriptomic data. Here's why it's so cool: - The MST connects static clinical data (tumor samples) in a way that minimizes the total distance between them, creating a "roadmap" of cancer progression. Combined with the appropriate dimensionality reduction and clustering approach this can be incredibly accurate. - Mahalanobis distance, unlike Euclidean distance, accounts for the covariance structure in our high-dimensional data, providing a more nuanced measure of similarity (and difference) between tumor profiles. - This approach allows us to reconstruct a pseudo-temporal ordering of tumor samples, revealing key insights into the evolving molecular functional landscape of endometrial cancer (and providing a model for future insights). It's been a learning experience how computational techniques can extract dynamic information from static data, offering new perspectives on cancer biology! I'm immensely grateful for the collaborative spirit of my team at Celcuity in enabling this study. Look forward to seeing how this research might contribute to data-driven cancer treatment strategies. Goes without saying - I always love thoughts and feedback from my LinkedIn colleagues - so reach out! #CancerResearch #Bioinformatics #DataScience #Collaboration #TrajectoryAnalysis #EndometrialCancer

19

1 Comment

Integrating multiomics and prior knowledge: a study of the Graphnet penalty impact https://lnkd.in/ewk6UnQz

22

Please check our latest publication in JITC: https://lnkd.in/dyjaKsfk In our recent study, we developed a novel small molecule library screen to identify compounds that enhance NK cell activity for cancer immunotherapy. We identified Cbl-b inhibitors as a key class of compounds that restored the effector functions of dysfunctional NK cells in tumor environments. Notably, Cbl-b inhibition rejuvenated NK cells from cancer patients, and when combined with TIGIT blockade, further boosted NK cell activity. These findings highlight the potential for combining Cbl-b inhibitors with checkpoint blockade therapies to enhance cancer immunotherapy. These results are the outcome of a collaborative effort between Roche and Alfred ZIppelius lab at University of Basel. A special thanks to all collaborators who contributed to this work! #Immunotherapy #NKCells #CancerResearch #JITC

56

7 Comments

In light of today's news about BioNTech's acquisition of Biotheus, I took a closer look at the clinical data concluding an encouraging antitumor activity and an acceptable safety profile of PM8002/BNT327. Here's a breakdown:  🔹 In locally advanced or metastatic TNBC - A phase Ib/II safety and efficacy study of PM8002/BNT327 in combination with nab-paclitaxel for first-line treatment: All pts received PM8002 and nab-paclitaxel. Primary objectives were safety and ORR per RECIST v1.1 by investigator assessment, with PFS and OS as the secondary objectives. Results (As of March 15, 2024): - 42 pts were enrolled. - Median duration of drug exposure: 10 months. - 13 pts were still on treatment. - Median PFS: 13.3 months for the ITT population. - ORR: 78.6% (including 1 complete response and 32 PRs) - cORR: 73.8% - DCR: 95.2%. - PD-L1 expression: . cORR in PD-L1 CPS<1: 76.9% . cORR in PD-L1 CPS≥1: 72.0% . And 100% PR in pts with PD-L1 CPS ≥10. - Adverse events: . All pts experienced TRAEs, 54.8% were Grade 3 or 4, no Grade 5 TRAEs. . 35.7% of pts experienced irAEs, 9.5% were Grade 3 or 4. . The most common AEs typically associated with VEGF inhibition were mostly Grade 1-2.  🔹 In EGFR-mutated NSCLC - A phase II safety and efficacy study of PM8002/BNT327 in combination with carboplatin and pemetrexed: Patients received PM8002/BNT327 plus carboplatin and pemetrexed Q3W for 4 cycles, followed by maintenance with PM8002/BNT327 and pemetrexed. The primary endpoint is ORR. Analysis by PD-L1 expression was determined by IHC with E1L3N clone; using TPS in biopsy taken after progression on EGFR-TKI therapy and classified based on TPS as negative (＜1%), low expression (1 to 49%), or high expression (≥50%). Results (As of April 12, 2024): - 64 patients were tested for PD-L1 TPS: 28 (43.8%) were TPS＜1%, 23 (35.9%) were TPS 1-49%, and 13 (20.3%) were TPS ≥50%. - Overall ORR: 54.7% (35/64, 95% CI:41.8-67.2) - DCR: 95.3% (61/64, 95% CI:86.9-99.0) - PD-L1 subgroups: . TPS＜1%: ORR 35.7% (95% CI:18.6-55.9), DCR 92.9% (95% CI:76.5-99.1) . TPS 1-49%: ORR 56.5% (95% CI:34.5-76.8), DCR 100% (95% CI:85.2-100.0). . TPS ≥50%: ORR 92.3% (95% CI:64.0-99.8), DCR 92.3% (95% CI:64.0-99.8). - Adverse events: . Any-grade TRAEs: 95.3% (61/64) . Grade ≥ 3 TRAEs: 54.7% (35/64). . Any-grade irAEs: 28.1% (18/64) . Grade ≥ 3 irAEs: 4.7% (3/64). . 6 patients discontinued PM8002 and/or chemotherapy due to TRAEs with 1 TRAE-related death.  🔹 In advanced RCC - A phase Ib/IIa safety and efficacy study of PM8002/BNT327 as a monotherapy. Due to character limitations in this post, please refer to the table below for efficacy results. #SolidTumors #OncologyDrugDiscovery #DualTargeting #PDL1 #VEGFa #BsAbs #BioNTech #Biotheus #BiotechAcquisition #OncologyRandD #BispecificAntibodies

10

1 Comment

‏ولله الحمد تم نشر بحثنا بعنوان: Sinefungin analogs targeting VP39 “ methyltransferase as potential anti‐monkeypox therapeutics: a multi‐step computational approach “ ‏بمجلة: ‏ “ Molecular Diversity “ ‏شكرا للدكتور Dr.Amr Salah Abouzied على مجهوداته العظيمة ولجميع الزملاء المشاركين في هذا البحث 🤍🙏🏻. It's my pleasure to share my recently published research article, titled "Sinefungin analogs targeting VP39 methyltransferase as potential anti‐monkeypox therapeutics: a multi‐step computational approach " in the "Molecular Diversity" 👉 https://lnkd.in/d5UxT8_u

152

23 Comments

In summary, to properly tackle age-related disease you need an ageing model. Most ageing models are incompatible with high throughput screening but C. elegans are small enough to fit into multiwell plates and age fast. Thanks to Chris Saunter and Robert Smith for input.

16

Oncology expertise is unevenly distributed therefore if LLM's could fill the gap that would have immediate consequences @NEJM_AI https://lnkd.in/eNC8jpmK 2044 unique questions x multiple LLMs. Useful checkpoint. Already looking forward to a Case Study with a 2025 update.

73

3 Comments

Researchers at USF Health and Weill Cornell Medicine, as part of an expansive, multi-institutional project investigating voice as a biomarker for disease, have reached a significant milestone by publishing the first version of their clinically validated voice dataset to an online artificial intelligence platform where it will be an invaluable resource for researchers across the globe. From our web story: "Artificial intelligence is revolutionizing our ability to detect and understand disease, and this groundbreaking voice dataset is a monumental step forward in that journey," said Dr. Olivier Elemento, Director of the Englander Institute for Precision Medicine, who is also a professor of physiology and biophysics at Weill Cornell Medicine. "These clinically validated data, combined with cutting-edge AI techniques, pave the way for new diagnostic possibilities and groundbreaking innovations that will transform patient care globally." #AI #PrecisionMedicine https://lnkd.in/esa69PqV

56

3 Comments

Digitalization may fundamentally change the practice of evidence generation in clinical trials. New Comment by Jörg Goldhahn Noe Brasier and Lindsay Kehoe from the Clinical Trials Transformation Initiative (CTTI) providing recommendations to unlock the full potential of digital health trials, including tools to develop digital biomarkers and endpoints, apply remote technology and interact with health authorities. https://lnkd.in/eiM-Jigt

40

2 Comments

Hot of the press! Preprint of new single cell method for targeted, sequence-specific capture of transcripts. Capture specific transcripts, transcript regions, fusion gene targets, mutations etc. together with whole transcriptome analysis. This method is only available with Rhapsody Single Cell analysis platform from BD. #singlecellanalysis #multiomics #RoCKseq https://lnkd.in/d4RiRWBv

44

1 Comment

How to differentiate hype from substance in the AI drug discovery space? A pretty cool take by Dr. Alex Zhavoronkov, co-founder and co-CEO of Insilico Medicine, published in Genetic Engineering & Biotechnology News today. Alex suggests to ask 10 simple but clear questions when evaluating AI drug discovery companies for the purpose of invstment or partnering: ✔ How many of your internally discovered drugs are in Phase I, Phase II, and Phase III? ✔ Did any of your pharma partnerships progress into Phase I, II, and III, and how long did it take? ✔ How many preclinical candidates (PCCs) did you nominate in a single year and in total? ✔ How long, on average, does it take you to nominate a PCC, and how much does it cost? ✔ What was the level of novelty of the targets and what is the level of novelty of the molecules? ✔ How long, on average, does it take you to go into clinical trials with an AI-discovered target? ✔ Did you in-license any of your drugs from other companies, repurpose them, or discover them internally? ✔ How many pharma companies use your AI software to discover drugs, and how many users do you have? How many renew every year? ✔ What was your revenue last year, and what was the revenue growth year over year? ✔ How much money have you raised, in how many rounds, and what is the valuation of your last round? What do you think? Write additional questions in the comments. I have my thoughts on it, but it is a topic for another post... #drugdiscovery #drugdesign #deeplearning Image credit: iStock

71

41 Comments