The discovery of new materials for separating ethylene from ethane by adsorption, instead of using cryogenic distillation, is a key milestone for molecular separations because of the multiple and widely extended uses of these molecules in industry. This technique has the potential to provide tremendous energy savings when compared with the currently used cryogenic distillation process for ethylene produced through steam cracking. Here we describe the synthesis and structural determination of a flexible pure silica zeolite (ITQ-55). This material can kinetically separate ethylene from ethane with an unprecedented selectivity of ~100, owing to its distinctive pore topology with large heart-shaped cages and framework flexibility. Control of such properties extends the boundaries for applicability of zeolites to challenging separations.

No abstract

Federated Learning is the current state of the art in supporting secure multi-party ML: data is maintained on the owner's device and is aggregated through a secure protocol [11]. However, this process assumes a trusted centralized infrastructure for coordination and clients must trust that the central service does not maliciously omit client contributions or use the byproducts of client data.As a response, we propose Biscotti: a fully decentralized P2P approach to multi-party ML, which uses blockchain and crypto primitives to coordinate a privacy-preserving ML process between peering clients. Our evaluation demonstrates that Biscotti is scalable, fault tolerant, and defends against known attacks. For example, Biscotti is able to protect the performance of the global model at scale even when 45% of adversaries are trying to poison the model [24].

No abstract

7006 Background: Measurable residual disease (MRD) prior to allogeneic hematopoietic cell transplantation (alloHCT) is associated with increased relapse and death in patients with acute myeloid leukemia (AML) in cytomorphological complete remission (CR). We recently demonstrated AML MRD detected in pre-alloHCT blood by DNA-sequencing was associated with increased relapse and decreased overall survival in patients randomized to reduced intensity conditioning (RIC) versus myeloablative conditioning (MAC). The clinical utility of such ultra-deep next-generation sequencing (NGS-MRD) had not yet been reported in a large multi-center cohort. Methods: Patients aged 18 or older who underwent first alloHCT between 2013-2017 for AML in first CR (CR1), reported to be FLT3, NPM1, IDH1, IDH2 and/or Kit mutated at diagnosis, with a pre-conditioning remission blood sample available in the CIBMTR biobank were eligible for this study. Ultra-deep anchored multiplex PCR-based NGS-MRD for the above mutations was performed on 500ng gDNA with error-corrected variant calling as previously described. The pre-specified statistical analysis plan was registered on OSF. Results: Of 457 patients with a sample available, 448 had sufficient clinical annotation and DNA for analysis. 147 of these 448 patients (33%) experienced relapse at a median of 5.6 months post-alloHCT. NGS-MRD was positive in 129 pre-alloHCT patient samples (29%), averaging 1.35 mutation(s)/patient (range: 1-4). 173 mutations were detected with a median VAF of 0.18% (range: 0.0054-62%), most frequently FLT3-ITD (n = 43 patients), NPM1 (n = 48), and IDH2 (n = 46). Testing positive by NGS-MRD prior to alloHCT was associated with a 3yr RFS of 36% (95% CI: 28-45%) compared with 56% (51-62%) in those testing negative (p < 0.001). Detection of NPM1 and/or FLT3-ITD mutations prior to alloHCT was associated with a 3yr relapse probability of 55% (43-67%) and RFS of 26% (16-37%). NGS-MRD impact was modified by conditioning intensity: positive patients receiving RIC/NMA had the highest relapse of 57% at 3yr, testing negative followed by RIC/NMA had the same relapse rate at 3yr (35%) as those who tested positive but received MAC (p < 0.001). At three years, those positive for FLT3-ITD and/or NPM1 mutations prior to RIC/NMA alloHCT had a relapse probability of 67% (50-83%) with a RFS of 19% (8-33%). HR for relapse if NGS-MRD positive pre-alloHCT in CR1 was 2.3 (p < 0.001, 95% CI 1.6-3.1) when adjusting for conditioning intensity and age group. Conclusions: In this largest cohort of NGS-MRD testing prior to alloHCT for AML reported to date, we confirm the ability to identify patients in CR1 but at high-risk of subsequent relapse. This evidence provides the foundation for future precision medicine approaches to reduce post-transplant AML relapse.