The role of complete lockdowns in reducing the reproduction ratios (Rt) of COVID-19 is now established. However, the persisting reality in many countries is no longer a complete lockdown, but restrictions of varying degrees using different choices of Non-pharmaceutical interaction (NPI) policies. A scientific basis for understanding the effectiveness of these graded NPI policies in reducing the Rt is urgently needed to address the concerns on personal liberties and economic activities. In this work, we develop a systematic relation between the degrees of NPIs implemented by the 26 cantons in Switzerland during March 9 to September 13 and their respective contributions to the Rt. Using a machine learning framework, we find that Rt which should ideally be lower than 1.0, has significant contributions in the post-lockdown scenario from the different activities - restaurants (0.0523 (CI. 0.0517-0.0528)), bars (0.030 (CI. 0.029-0.030)), and nightclubs (0.154 (CI. 0.154-0.156)). Activities which keep the land-borders open (0.177 (CI. 0.175-0.178)), and tourism related activities contributed comparably 0.177 (CI. 0.175-0.178). However, international flights with a quarantine did not add further to the Rt of the cantons. The requirement of masks in public transport and secondary schools contributed to an overall 0.025 (CI. 0.018-0.030) reduction in Rt, compared to the baseline usage even when there are no mandates. Although causal relations are not guaranteed by the model framework, it nevertheless provides a fine-grained justification for the relative merits of choice and the degree of the NPIs and a data-driven strategy for mitigating Rt.

Several questions resonate as the governments relax their COVID-19 mitigation policies - is it too early to relax them, were the policies as effective as they could have been. Answering these questions about the past or crafting newer policy decisions in the future requires a quantification of how policy choices affect the spread of the infection. Policy landscape as well as the infection trajectories from different states and countries diverged so fast that comparing and learning from them has not been easy. In this work, we standardize and pool together the ensemble of lockdown and graded re-opening policies adopted by the 50 states of USA in any given week between 9th March and 9th August. Using artificial intelligence (AI) on this pooled data, we build a predictive model (R2training=0.79, R2test=0.76) for the weekly-averaged transmission rate of infections. Predictability conceptually raises the possibility of an evidence-based or data-driven mitigation policy-making by evaluating the relative merits of the different policy scenarios. Probing the predictions with interpretable AI highlights how factors such as the closing of bars or the use of masks influence transmission, effects which have been hard to decouple from the ensemble of policy instrument combinations. While acknowledging the limitations of our predictions as well as of the infection testing, we ask the theoretical question if the observed transmission rates in the states were as efficient as they could have been under various levels of restrictions, and if the mitigation policies of the states are overdesigned. The model can be further refined with a more detailed inclusion of geographies and policy compliances, as well as expanded as newer policies emerge.

Unconventional protein secretion (UPS) is an important phenomenon with fundamental implications to cargo export. How eukaryotic proteins transported by UPS are recognized without a conventional signal peptide has been an open question. It was recently observed that a diacidic amino acid motif (ASP-GLU or DE) is necessary for the secretion of superoxide dismutase 1 (SOD1) from yeast under nutrient starvation. Taking cue from this discovery, we explore the hypothesis of whether the diacidic motif DE, which can occur fairly ubiquitously, along with its context, can be a generic signal for unconventional secretion of proteins. Four different contexts were evaluated: a physical context encompassing the structural order and charge signature in the neighbourhood of DE, two signalling contexts reflecting the presence of either a phosphorylatable amino acid (‘X’ in XDE, DXE, DEX) or an LC3 interacting region (LIR) which can trigger autophagy and a co-evolutionary constraint relative to other amino acids in the protein interpreted by examining sequences across different species. Among the 100 proteins we curated from different physiological or pathological conditions, we observe a pattern in the unconventional secretion of heat shock proteins in the cancer secretome, where DE in an ordered structural region has higher odds of being a UPS signal.

Acinetobacter baumannii, has been developing resistance to even the last line of drugs. Antimicrobial peptides (AMPs) to which bacteria do not develop resistance easily may be the last hope. A few independent experimental studies have designed and studied the activity of AMPs on A. baumannii, however the number of such studies are still limited. With the goal of developing a rational approach to the screening of AMPs against A. baumannii, we carefully curated the drug activity data from 75 cationic AMPs, all measured with a similar protocol, and on the same ATCC 19606 strain. A quantitative model developed and validated with a part of the data. While the model may be used for predicting the activity of any designed AMPs, in this work, we perform an in silico screening for the entire database of naturally occurring AMPs, to provide a rational guidance in this urgently needed drug development.

Rational design methodologies such as quantitative structure activity relationships (QSAR) have conventionally focused on screening through several drugs for their activity against a single target, either a bacterial protein or membrane. Recent concerns in drug design such as the development of drug resistance by membrane adaptation, or the undesirable damage to gut microbiota require a paradigm shift in activity prediction.A complementary approach capable of predicting the activity of a single drug against diverse targets, the diversity arising from bacterial adaptation or a heterogeneous composition with other helpful or harmful bacteria, is needed. As a first predictive step towards this goal, we develop a quantitative model for the activity of daptomycin on Streptococcus aureus strains with different membrane compositions, mainly varying in lysylation. The results of the predictions are good, and within the limits of the scarcely 1 available data, hint at an interaction of daptomycin with the inner membrane. The complementary approach may in principle be extended to estimate the activity against gut bacterial membranes, when systematic data can be curated for training the model.