Exploring the Structural Attributes of Yoda1 for the Development of New-Generation Piezo1 Agonist Yaddle1 as a Vaccine Adjuvant Targeting Optimal T Cell Activation.

Piezo1, a mechano-activated ion channel, has wide-ranging physiological and therapeutic implications, with the ongoing development of specific agonists unveiling cellular responses to mechanical stimuli. In our study, we systematically analyzed the chemical subunits in Piezo1 protein agonist Yoda1 to comprehend the structure-activity relationship and push forward next-generation agonist development. Preliminary screening assays for Piezo1 agonism were performed using the Piezo1-mCherry-transfected HEK293A cell line, keeping Yoda1 as a positive control. We introduce a novel Piezo1 agonist Yaddle1 (34, 0.40 µM), featuring a trifluoromethyl group, with further exploration through in vitro studies and density functional theory calculations, emphasizing its tetrel interactions, to act as an ambidextrous wedge between the domains of Piezo1. In contrast to the poor solubility of the established agonist Yoda1, our results showed that the kinetic solubility of Yaddle1 (26.72 ± 1.8 µM at pH 7.4) is 10-fold better than that of Yoda1 (1.22 ± 0.11 µM at pH 7.4). Yaddle1 (34) induces Ca2+ influx in human CD4+ T cell, suggesting its potential as a vaccine adjuvant for enhanced T cell activation.

Mechanistic Inquisition on the Reduction of C17Si(NH2)2 to NH3: A DFT Study.

Activation of molecular nitrogen by silicon-substituted cyclo[18]carbon and its ability to produce the C17Si-(NH2)2 derivative, as the precursor of NH3, has been recently reported. This specific acquisition has piqued an interest to investigate the possibility of NH3 formation with further addition of H2 molecules in the gaseous reaction media. The current investigations reported in this article show that two moles of molecular H2 generate two molecules of NH3 and a C17Si-H2 byproduct from its precursor. The catalyst gets restored by an in situ reaction between some unreacted C17Si-N2 and the byproduct in the media. This reaction also produces the next C17Si-(NH)2 adduct, which restarts the catalytic cycle for NH3 production again.

Unraveling the origin of the cooperative adsorption of carbon monoxide in an Fe(II) metal-organic framework.

Using periodic DFT calculations, we have established the mechanism of the unusual cooperative adsorption of CO gas in an Fe-bistriazolate MOF observed previously. The binding of one CO molecule to FeII triggers structural alteration of the neighbouring Fe centres, reducing the steric energy penalty and aiding cooperative adsorption. This is similar to the entatic state concept proposed for metalloenzymes, and offers novel strategies for selective gas adsorption.

Synchronous Proton-Hydride Transfer by a Pyrazole-Functionalized Protic Mn(I) Complex in Catalytic Alcohol Dehydrogenative Coupling.

A series of Mn(I) complexes Mn(L1 )(CO)3 Br, Mn(L2 )(CO)3 Br, Mn(L1 )(CO)3 (OAc) and Mn(L3 )(CO)3 Br [L1 =2-(5-tert-butyl-1H-pyrazol-3-yl)-1,8-naphthyridine, L2 =2-(5-tert-butyl-1H-pyrazol-3-yl)pyridine, L3 =2-(5-tert-butyl-1-methyl-1H-pyrazol-3-yl)-1,8-naphthyridine] were synthesized and fully characterized. The acid-base equilibrium between the pyrazole and the pyrazolato forms of Mn(L1 )(CO)3 Br was studied by 1 H NMR and UV-vis spectra. These complexes are screened as catalysts for acceptorless dehydrogenative coupling (ADC) of primary alcohols and aromatic diamines for the synthesis of benzimidazole and quinoline derivatives with the release of H2 and H2 O as byproducts. The protic complex Mn(L1 )(CO)3 Br shows the highest catalytic activity for the synthesis of 2-substituted benzimidazole derivatives with broad substrate scope, whereas a related complex [Mn(L3 )(CO)3 Br], which is devoid of the proton responsive ß-NH unit, shows significantly reduced catalytic efficiency validating the crucial role of the ß-NH functionality for the alcohol dehydrogenation reactions. Control experiments, kinetic and deuterated studies, and density functional theory (DFT) calculations reveal a synchronous hydride-proton transfer by the metal-ligand construct in the alcohol dehydrogenation step.

Sheehan's Syndrome unmasked by dengue fever: A case report and review of literature.

Sheehan's syndrome is a pituitary disease resulting from severe postpartum hemorrhage and can present with varying degrees of pituitary insufficiency. Although its incidence is decreasing in developed countries, it continues to be one of the most common causes of hypopituitarism in underdeveloped and developing countries. Here, we report a case of Sheehan's syndrome which was diagnosed following an episode of severe dengue infection, in a 38-year-old female.

Monolayer molybdenum diborides containing flat and buckled boride layers as anode materials for lithium-ion batteries.

The materials community is interested in discovering new two-dimensional (2D) crystals because of the potential for fascinating features. In this work, by employing a systematic first-principles DFT analysis and MD simulations, we investigated the potential applications of monolayer Mo borides containing flat and buckled boride rings named P6/mmm and R3̄m MoB2 as anode materials of lithium-ion batteries. Our preliminary investigations show that the MoB2 monolayers possess significant structural, thermodynamic, mechanical, and dynamical stability. Due to their distinctive crystal structures, the Mo borides exhibit unique electronic properties, as expected. Additionally, we discovered that the highly negative Li adsorption energy achieved can aid in stabilizing the Li's adsorption on the surface of MoB2 rather than clustering, which ensured its suitability for LIB anode applications. The low computed Li-ion and Li-vacancy migration energy barrier provides robust charge/discharge performance even at a fully lithiated state, signifying their extraordinary possibility of being a suitable anode material for Li batteries. Both the monolayers can hold a maximum of two layers of Li ions on both sides to give a huge specific capacity of 912 mA h g-1, much higher than graphene and MoS2-based anodes. The computed in-plane stiffness constants demonstrate that the monolayer pristine and lithiated MoB2 satisfies Born's criteria, implying its mechanical flexibility. Additionally, its strong mechanical and thermal properties at the pristine and the lithiated state indicate that the 2D MoB2 can withstand massive volume expansion at a high temperature of 500 K during the lithiation/de-lithiation reaction and is remarkably beneficial for manufacturing flexible anodes. Based on the above findings, these two newly designed classes of monolayers of MoB2 are anticipated to open a new avenue for the upcoming generation of lithium-ion batteries.

Evaluation of Rapid Antigen Test as a Marker of SARS-CoV-2 Infectivity.

Background Coronavirus disease 2019 (COVID-19) caused by severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) that emerged in late 2019 continues to spread globally. Reverse transcriptase polymerase chain reaction (RT-PCR), which is considered the gold standard for diagnosis, does not always indicate contagiousness. This study was planned to evaluate the performance of the rapid antigen test (RAT) with the duration of symptoms and the usefulness of these tests in determining the infectivity of patients by performing sub-genomic RT-PCR. Methodology This prospective, observational study was designed to compare the diagnostic value of the COVID-19 RAT (SD Biosensor, Korea) with COVID-19 RT-PCR (Thermo Fisher, USA) by serial testing of patients. To evaluate the infectivity of the virus, sub-genomic RT-PCR was performed on previous RAT and RT-PCR-positive samples. Results Of 200 patients, 102 were positive on both RT-PCR and RAT, with 87 patients serially followed and tested. The sensitivity and specificity of RAT were 92.73% and 93.33%, respectively, in symptomatic patients. The mean duration of RAT positivity was 9.1 days, and the mean duration of RT-PCR positivity was 12.6 days. Sub-genomic RT-PCR test was performed on samples that were reported to be positive by RAT, and 73/87 (83.9%) patients were found to be positive. RAT was positive in symptomatic patients whose duration of illness was less than 10 days or those with a cycle threshold value below 32. Conclusions Thus, RAT can be used as the marker of infectivity of SARS-CoV-2 in symptomatic patients, especially in healthcare workers.

Cytomegalovirus reactivation with high viral load in a patient of coronavirus disease 2019 acute respiratory distress syndrome: a case report.

INTRODUCTION: Cytomegalovirus establishes life-long latency after primary infection in childhood. Cytomegalovirus reactivation has been well reported in immune-compromised patients; however, in the last few years it has been observed that cytomegalovirus reactivation also occurs in critically ill patients without exogenous immunosuppression, which increases length of intensive care unit stay and mortality rate. CASE REPORT: A 63-year-old Indian male, without any known comorbidity, developed severe coronavirus disease 2019 and was admitted to the intensive care unit. He received remdesivir, tocilizumab, steroids, anticoagulants, and empiric antibiotics over the next 3 weeks. However, his clinical condition did not improve much, and during the 9th week of illness his condition started deteriorating and routine bacterial cultures, fungal cultures, and cytomegalovirus real-time polymerase chain reaction on blood were negative. His clinical condition worsened rapidly, which led to the need for invasive mechanical ventilation. Tracheal aspirate bacterial and fungal culture showed no growth, but cytomegalovirus real-time polymerase chain reaction showed 21,86,000 copies/mL in tracheal aspirates. After 4 weeks of ganciclovir treatment, the patient improved clinically and was discharged. Currently he is doing well and able to do his routine activity without the need of oxygen. CONCLUSION: Timely management with ganciclovir is associated with favorable outcome in cytomegalovirus infection. Thus, it can be suggested that treatment should be initiated with ganciclovir if a patient with coronavirus disease 2019 has high cytomegalovirus load in tracheal aspirates, along with unexplained and prolonged clinical and/or radiological features.

Role of molecular modelling in the development of metal-organic framework for gas adsorption applications.

More than 47,000 articles have been published in the area of Metal-Organic Framework since its seminal discovery in 1995, exemplifying the intense research carried out in this short span of time. Among other applications, gas adsorption and storage are perceived as central to the MOFs research, and more than 10,000 MOFs structures are reported to date to utilize them for various gas storage/separation applications. Molecular modeling, particularly based on density functional theory, played a key role in (i) understanding the nature of interactions between the gas and the MOFs geometry (ii) establishing various binding pockets and relative binding energies, and (iii) offering design clues to improve the gas uptake capacity of existing MOF architectures. In this review, we have looked at various MOFs that are studied thoroughly using DFT/periodic DFT (pDFT) methods for CO2, H2, O2, and CH4 gases to provide a birds-eye-view on how various exchange-correlation functionals perform in estimating the binding energy for various gases and how factors such as nature of the (i) metal ion, (ii) linkers, (iii) ligand, (iv) spin state and (v) spin-couplings play a role in this process with selected examples. While there is still room for improvement, the rewards offered by the molecular modelling of MOFs were already substantial that we advocate experimental and theoretical studies to go hand-in-hand to undercut the trial-and-error approach that is often perceived in the selection of MOFs and gas partners in this area. Graphical abstract: The importance of density functional theory-based molecular modeling studies in offering design clues to improve the gas adsorption and storage capacity of existing MOF architectures is discussed here. The use of DFT-based investigation in conjunction with experimental synthesis is an imperative tool in designing new-generation MOFs with enhanced uptake capacity.

Emergence of Unusual Microorganisms in Microflora of Pilonidal Sinuses: A Multiple Case Series.

PURPOSE: Recent reports have noted an emergence of unusual organisms in microflora of pilonidal sinus (PNS); this study was undertaken to identify the primary microbial flora associated with infected primary PNS over a period of 1 year. DESIGN: A prospective multiple case series. SUBJECTS AND SETTING: A case series of 20 patients with primary PNS in a tertiary care center in Varanasi, India, was studied. The study was conducted at the Department of Microbiology and General Surgery, Institute of Medical Sciences, Varanasi, over a period of 1 year (September 2016 to July 2017). METHODS: Purulent exudate (pus) samples were collected from 20 patients with primary PNS from the discharging sinuses by aseptic methods. Samples were assessed for aerobic and anaerobic flora by conventional culture and molecular methods. Antimicrobial susceptibility testing was done for bacterial isolates. Bacterial diversity was compared with the demographic and clinical profile of the sinuses by multiple correspondence analysis. RESULTS: Of the total cases, 11 (55%) had purulent discharge, among which all showed polymicrobial flora. The ratio of aerobic to anaerobic organisms was 1:2 (16/32). Escherichia coli (E. coli, 4, 36.36%) and Enterococcus faecalis (E. faecalis, 4, 36.36%) were commonly isolated. Bifidobacterium was the most frequent anaerobe. Detailed molecular analysis revealed the presence of Kocuria flava as an unusual pathogen. On statistical analysis, factors like male gender, increased body mass index, absence of hair in sinus, presence of features of hirsutism, and absence of Fusobacteria were closely associated with one another in these PNS cases. CONCLUSIONS: The case series revealed the predominance of anaerobes in primarily infected PNS cases. Bifidobacterium spp and unusual pathogens like K. flava were among the emerging pathogens in infected PNS. Use of better molecular diagnostic facilities in addition to the conventional methods might enhance the verified diversity of microorganisms in such cases.

A Theoretical Perspective to Decipher the Origin of High Hydrogen Storage Capacity in Mn(II) Metal-Organic Framework.

Herein, we report a detailed periodic DFT investigation of Mn(II)-based [(Mn4 Cl)3 (BTT)8 ]3- (BTT3- =1,3,5-benzenetristetrazolate) metal-organic framework (MOF) to explore various hydrogen binding pockets, nature of MOF…H2 interactions, magnetic coupling and, H2 uptake capacity. Earlier experiments found an uptake capacity of 6.9 wt % of H2, with the heat of adsorption estimated to be ∼10 kJ/mol, which is one among the highest for any MOFs reported. Our calculations unveil different binding sites with computed binding energy varying from -6 to -15 kJ/mol. The binding of H2 at the Mn2+ site is found to be the strongest (site I), with H2 found to bind Mn2+ ion in a η2 fashion with a distance of 2.27 Šand binding energy of -15.4 kJ/mol. The bonding analysis performed using NBO and AIM reveal a strong donation of σ (H2 ) to the dz 2 orbital of the Mn2+ ion responsible for such large binding energy. The other binding pockets, such as -Cl (site II) and BTT ligands (site III and IV) were found to be weaker, with the binding energy decreasing in the order I>II>III>IV. The average binding energy computed for these four sites put together is 9.6 kJ/mol, which is in excellent agreement with the experimental value of ∼10 kJ/mol. We have expanded our calculations to compute binding energy for multiple sites simultaneously, and in this model, the binding energy per site was found to decrease as we increased the number of H2 molecules suggesting electronic and steric factors controlling the overall uptake capacity. The calculated adsorption isotherm using the GCMC method reproduces the experimental observations. Further, the magnetic coupling computed for the unbound MOF reveals moderate ferromagnetic and strong antiferromagnetic coupling within the tetrameric {Mn4 } unit leading to a three-up-one-down spin configuration as the ground state. These were then coupled ferromagnetically to other tetrameric units in the MOF network. The magnetic coupling was found to alter only marginally upon gas binding, suggesting that both exchange interaction and the spin-states are unlikely to play a role in the H2 uptake. This is contrary to the O2 uptake studied lately, where strong dependence on exchange-coupling/spin state was witnessed, suggesting exchange-coupling/magnetic field dependent binding as a viable route for gas separation.

Activation and Conversion of Molecular Nitrogen to the Precursor of Ammonia on Silicon Substituted Cyclo[18]Carbon: a DFT Design.

Recent synthesis of sp-hybridized cyclo[18]carbon allotrope has attracted immense curiosity. Since then, a generous amount of theoretical studies concerning aromaticity, adsorption, and spectra of the molecule have been performed. However, very few stuides have been carried out concerning its reactivities and catalytic behaviour. In this article, a DFT-based inquisition has been reported regarding the reactivity of Si substituted cyclo[18]carbon molecule towards molecular N2 . Results show that the Si substituted derivative is effective in producing adducts with molecular nitrogen. Charge calculations and IRC trapping methods indicate that only the Si center of C17 Si and its (HOMO-1) level participate in N2 addition. The N-adduct so formed, is then found to spontaneously react with molecular H2 . The addition of two H2 molecules to the activated nitrogen molecule to give respective amine derivatives have also been studied. The successful generation of the precursor of NH3 by C17 Si lays a clear emphasis on its potentiality.

Development, Optimization, and In Vivo Validation of New Imidazopyridine Chemotypes as Dual TLR7/TLR9 Antagonists through Activity-Directed Sequential Incorporation of Relevant Structural Subunits.

Undesirable activation of endosomal toll-like receptors TLR7 and TLR9 present in specific immune cells in response to host-derived ligands is implicated in several autoimmune diseases and other contexts of autoreactive inflammation, making them important therapeutic targets. We report a drug development strategy identifying a new chemotype for incorporating relevant structural subunits into the basic imidazopyridine core deemed necessary for potent TLR7 and TLR9 dual antagonism. We established minimal pharmacophoric features in the core followed by hit-to-lead optimization, guided by in vitro and in vivo biological assays and ADME. A ligand-receptor binding hypothesis was proposed, and selectivity studies against TLR8 were performed. Oral absorption and efficacy of lead candidate 42 were established through favorable in vitro pharmacokinetics and in vivo pharmacodynamic studies, with IC50 values of 0.04 and 0.47 µM against TLR9 and TLR7, respectively. The study establishes imidazopyridine as a viable chemotype to therapeutically target TLR9 and TLR7 in relevant clinical contexts.

Hydroalcoholic Extract of Sechium edule Fruits Attenuates QT Prolongation in High Fat Diet-Induced Hyperlipidemic Mice.

The present study aimed to evaluate the effect of hydroalcoholic extract of Sechium edule (S.E.) fruits on lipid profile and electrocardiogram (ECG) parameters in high fat-diet (HFD) induced hyperlipidemic mice. In this study, grouping of animals was done as described below (n = 6), where group 1 is normal control, group 2 is HFD control, group 3 is HFD + atorvastatin (10 mg/kg), group 4 is HFD + S.E. extract (200 mg/kg), and group 5 is HFD + S.E. extract (400 mg/kg). The first 3 weeks animals were supplemented with HFD, and the last 3 weeks animals were supplemented with HFD along with atorvastatin (10 mg/kg) or S.E. extract (200 and 400 mg/kg). It was observed that mice of the HFD control group showed a significant rise in the total cholesterol, triglycerides, LDL-C, and VLDL-C levels and a notable decrease in HDL-C levels. In addition, a consequential increment in ECG parameters such as QT or QTc and RR interval and a noteworthy decline in the heart rate were observed in HFD control mice. Treatment with S.E. extract (200 and 400 mg/kg) showed a significant improvement in the lipid profile. Moreover, the extract also significantly normalized the prolonged QT or QTc and RR interval and the heart rate in HFD-challenged mice. Hence, we can conclude that S.E. extract encumbers the prolongation of QT or QTc and RR interval and increased the heart rate in HFD-challenged mice.

Integration of Ligand-Based and Structure-Based Methods for the Design of Small-Molecule TLR7 Antagonists.

Toll-like receptor 7 (TLR7) is activated in response to the binding of single-stranded RNA. Its over-activation has been implicated in several autoimmune disorders, and thus, it is an established therapeutic target in such circumstances. TLR7 small-molecule antagonists are not yet available for therapeutic use. We conducted a ligand-based drug design of new TLR7 antagonists through a concerted effort encompassing 2D-QSAR, 3D-QSAR, and pharmacophore modelling of 54 reported TLR7 antagonists. The developed 2D-QSAR model depicted an excellent correlation coefficient (R2training: 0.86 and R2test: 0.78) between the experimental and estimated activities. The ligand-based drug design approach utilizing the 3D-QSAR model (R2training: 0.95 and R2test: 0.84) demonstrated a significant contribution of electrostatic potential and steric fields towards the TLR7 antagonism. This consolidated approach, along with a pharmacophore model with high correlation (Rtraining: 0.94 and Rtest: 0.92), was used to design quinazoline-core-based hTLR7 antagonists. Subsequently, the newly designed molecules were subjected to molecular docking onto the previously proposed binding model and a molecular dynamics study for a better understanding of their binding pattern. The toxicity profiles and drug-likeness characteristics of the designed compounds were evaluated with in silico ADMET predictions. This ligand-based study contributes towards a better understanding of lead optimization and the future development of potent TLR7 antagonists.

Unraveling the Mechanistic Details of Ru-Bis(pyridyl)borate Complex Catalyst for the Dehydrogenation of Ammonia Borane.

Ru-Bis(pyridyl)borate complex (CAT) is an efficient catalyst for ammonia borane (AB) dehydrogenation. Although the mechanistic pathway of this catalyst has been theoretically investigated previously, the gap between the experimental findings and the computational results could not be bridged thus far. In our study, using density functional theory calculations, we elucidate the mechanism of AB dehydrogenation of CAT at a variable degree of ligand hydrogenation. Our results confirm that the acetonitrile ligands get reduced in the presence of AB and remain hydrogenated. Moreover, in line with experiments, we find that AB dehydrogenation on CAT proceeds via a concerted mechanism (with the free energy energetic span between 25.4 and 32.5 kcal/mol). We find that the ligand reduction alters the electronic structure and activity of CAT and the highest activity of the catalyst is expected at the fifth degree of hydrogenation of ligands with an energetic span of 25.4 kcal/mol. Additionally, the mechanism for the removal of molecular H2 from the catalysts also alters with the degree of ligand hydrogenation. Furthermore, our results show that optimal H2 binding free energy calculations can be used as a descriptor to identify the most active sites. Finally, this work demonstrates that ligand reduction improves the activity of the catalyst. These results highlight the importance of ligand hydrogenation in probing the activity and operating mechanism of the Ru-bis(pyridyl)borate complexes for AB dehydrogenation. Further, we identify a plausible dimer structure and rationalized experimental observation that the deactivation chemistry of this catalyst is different from the Shvo's catalyst.

Influence of Linker Orientation and Regulative Factor(s) in Liposomal Gene Delivery: A Molecular Level Investigation.

Molecular level understanding of liposome-gene interaction is immensely important for the research progress and technological advancement of gene delivery, which is highly significant due to a wide range of applications of gene therapy. The liposomal gene delivery method is one of the most promising techniques due to its efficacy to easily fuse with the cell membrane and its lower toxicity. In vivo gene delivery using liposomes is reported to be extremely successful. However, the success of gene delivery depends on various factors including the chemical nature of the structural unit of the liposome. To explore the regulative factor(s) for liposomal gene delivery, we systematically analyze the linker orientation effect on the gene delivery efficiency of liposomes through a density functional theory (DFT) study. Interestingly, it is observed that the liposome-gene interaction is not the regulating factor for successful gene delivery. The success depends on the gel to liquid melting temperature of the liposome.

The Decisive Role of Spin States and Spin Coupling in Dictating Selective O2 Adsorption in Chromium(II) Metal-Organic Frameworks.

Invited for the cover of this issue are Sourav Pal, Gopalan Rajaraman and co-workers at the Indian Institute of Technology Bombay, the Bhabha Atomic Research Centre and the Indian Institute of Science Education and Research. The image depicts how a mixture of atmospheric gases such as CO2 , H2 , N2 and O2 can be selectively separated using a Cr metal-organic framework where spin state and spin coupling play a crucial role. Read the full text of the article at 10.1002/chem.202104526.

The Decisive Role of Spin States and Spin Coupling in Dictating Selective O2 Adsorption in Chromium(II) Metal-Organic Frameworks.

The coordinatively unsaturated chromium(II)-based Cr3 [(Cr4 Cl)3 (BTT)8 ]2 (Cr-BTT; BTT3- =1,3,5-benzenetristetrazolate) metal-organic framework (MOF) has been shown to exhibit exceptional selectivity towards adsorption of O2 over N2 /H2 . Using periodic density functional theory (DFT) calculations, we attempted to decipher the origin of this puzzling selectivity. By computing and analyzing the magnetic exchange coupling, binding energies, the partial density of states (pDOS), and adsorption isotherms for the pristine and gas-bound MOFs [(Cr4 (X)4 Cl)3 (BTT)8 ]3- (X=O2 , N2 , and H2 ), we unequivocally established the role of spin states and spin coupling in controlling the gas selectivity. The computed geometries and gas adsorption isotherms are consistent with the earlier experiments. The binding of O2 to the MOF follows an electron-transfer mechanism resulting in a CrIII superoxo species (O2 .- ) with a very strong antiferromagnetic coupling between the two centers, whereas N2 /H2 are found to weakly interact with the metal center and hence only slightly perturb the associated coupling constants. Although the gas-bound and unbound MOFs have an S=0 ground state (GS), the nature of spin the configurations and the associated magnetic exchanges are dramatically different. The binding energy and the number of oxygen molecules that can favorably bind to the Cr center were found to vary with respect to the spin state, with a significant energy margin (47.6 kJ mol-1 ). This study offers a hitherto unknown strategy of using spin state/spin couplings to control gas adsorption selectivity in MOFs.

Deep Unlearning via Randomized Conditionally Independent Hessians.

Recent legislation has led to interest in machine unlearning, i.e., removing specific training samples from a predictive model as if they never existed in the training dataset. Unlearning may also be required due to corrupted/adversarial data or simply a user's updated privacy requirement. For models which require no training (k-NN), simply deleting the closest original sample can be effective. But this idea is inapplicable to models which learn richer representations. Recent ideas leveraging optimization-based updates scale poorly with the model dimension d, due to inverting the Hessian of the loss function. We use a variant of a new conditional independence coefficient, L-CODEC, to identify a subset of the model parameters with the most semantic overlap on an individual sample level. Our approach completely avoids the need to invert a (possibly) huge matrix. By utilizing a Markov blanket selection, we premise that L-CODEC is also suitable for deep unlearning, as well as other applications in vision. Compared to alternatives, L-CODEC makes approximate unlearning possible in settings that would otherwise be infeasible, including vision models used for face recognition, person re-identification and NLP models that may require unlearning samples identified for exclusion. Code is available at https://github.com/vsingh-group/LCODEC-deep-unlearning.