**Abstract:**

Migration has played a pivotal role in shaping both cultural and genetic diversity throughout human history. Results from population genetics show that even low rates of migration between populations lead to the homogenization of their genetic compositions. On the cultural side, we know that conformity helps maintain differences between groups. However, equivalent insights into the effect of migration on cultural composition remain limited. To address this gap, I have developed a simulation model that explores the joint evolution of genetic and cultural traits, investigating how varying strengths of conformity and migration rates influence differentiation both within and between populations

**Speaker's social media:**

Twitter: @camilla_cenni

LIVE - Cultural drivers of stone-tool use in Balinese long-tailed macaques

Event extraction (EE), as a crucial information extraction (IE) task, aims to identify event triggers and their associated arguments from unstructured text, subsequently classifying them into pre-defined types and roles. In the biomedical domain, EE is widely used to extract complex structures representing biological events from literature. Due to the complicated semantics and specialized domain knowledge, it is challenging to construct biomedical event extraction datasets. Additionally, most existing biomedical EE datasets primarily focus on cell experiments or the overall experimental procedures. Therefore, we introduce AniEE, an event extraction dataset concentrated on the animal experiment stage. We establish a novel animal experiment customized entity and event scheme in collaboration with domain experts. We then create an expert-annotated high-quality dataset containing discontinuous entities and nested events and evaluate our dataset on the recent outstanding NER and EE models.

AniEE: A Dataset of Animal Experimental Literature for Event Extraction

Recently, improving the robustness of policies across different environments attracts increasing attention in the reinforcement learning (RL) community. Existing robust RL methods mostly aim to achieve the max-min robustness by optimizing the policy’s performance in the worst-case environment. However, in practice, a user that uses an RL policy may have different preferences over its performance across environments. Clearly, the aforementioned max-min robustness is oftentimes too conservative to satisfy user preference. Therefore, in this paper, we integrate user preference into policy learning in robust RL, and propose a novel User-Oriented Robust RL (UOR-RL) framework. Specifically, we define a new User-Oriented Robustness (UOR) metric for RL, which allocates different weights to the environments according to user preference and generalizes the max-min robustness metric. To optimize the UOR metric, we develop two different UOR-RL training algorithms for the scenarios with or without a priori known environment distribution, respectively. Theoretically, we prove that our UOR-RL training algorithms converge to near-optimal policies even with inaccurate or completely no knowledge about the environment distribution. Furthermore, we carry out extensive experimental evaluations in 6 MuJoCo tasks. The experimental results demonstrate that UOR-RL is comparable to the state-of-the-art baselines under the average-case and worst-case performance metrics, and more importantly establishes new state-of-the-art performance under the UOR metric.

User-Oriented Robust Reinforcement Learning

Safety comes first in many real-world applications involving autonomous agents. Despite a large number of reinforcement learning (RL) methods focusing on safety-critical tasks, there is still a lack of high-quality evaluation of those algorithms that adheres to safety constraints at each decision step under complex and unknown dynamics. In this paper, we revisit prior work in this scope from the perspective of state-wise safe RL and categorize them as projection-based, recovery-based, and optimization-based approaches, respectively.  Furthermore, we propose Unrolling Safety Layer (USL), a joint method that combines safety optimization and safety projection. This novel technique explicitly enforces hard constraints via the deep unrolling architecture and enjoys structural advantages in navigating the trade-off between reward improvement and constraint satisfaction. To facilitate further research in this area,  we reproduce related algorithms in a unified pipeline and incorporate them into SafeRL-Kit, a toolkit that provides off-the-shelf interfaces and evaluation utilities for safety-critical tasks. We then perform a comparative study of the involved algorithms on six benchmarks ranging from robotic control to autonomous driving. The empirical results provide an insight into their applicability and robustness in learning zero-cost-return policies without task-dependent handcrafting. The project page is available at \url{https://sites.google.com/view/saferlkit}.

Evaluating Model-free Reinforcement Learning toward Safety-critical Tasks

Cooperative multi-agent reinforcement learning (c-MARL) is widely applied in safety-critical scenarios, thus the analysis of robustness for c-MARL models is profoundly important. However, robustness certification for c-MARLs has not yet been explored in the community. In this paper, we propose a novel certification method, which is the first work to leverage a scalable approach for c-MARLs to determine actions with guaranteed certified bounds. c-MARL certification poses two key challenges compared with single-agent systems: (i) the accumulated uncertainty as the number of agents increases; (ii) the potential lack of impact when changing the action of a single agent into a global team reward. These challenges prevent us from directly using existing algorithms. Hence, we employ the false discovery rate (FDR) controlling procedure considering the importance of each agent to certify per-state robustness and propose a tree-search-based algorithm to find a lower bound of the global reward under the minimal certified perturbation. As our method is general, it can also be applied in single-agent environments. We empirically show that our certification bounds are much tighter than those of the state-of-the-art RL certification solutions. We also run experiments on two popular c-MARL algorithms: QMIX and VDN, in two different environments, with two and four agents. The experimental results show that our method produces a meaningful guaranteed robustness for all models and environments. Our tool CertifyCMARL is available at https://github.com/TrustAI/CertifyCMARL.

Certified Policy Smoothing for Cooperative Multi-Agent Reinforcement Learning

As the complexity of pacemaker devices continues to grow, the importance of capturing its functional correctness requirement formally cannot be overestimated. The pacemaker system specification document by \emph{Boston Scientific} provides a widely accepted set of specifications for pacemakers. 
As these specifications are written in a natural language, they are not amenable for automated verification, synthesis, or reinforcement learning of pacemaker systems. This paper presents a formalization of these requirements for a dual-chamber pacemaker in \emph{duration calculus} (DC), a highly expressive real-time specification language.
The proposed formalization allows us to automatically translate pacemaker requirements into executable specifications as stopwatch automata, which can be used to enable simulation, monitoring, validation, verification and automatic synthesis of pacemaker systems. 
The cyclic nature of the pacemaker-heart closed-loop system results in DC requirements that compile to a decidable subclass of stopwatch automata. We present shield reinforcement learning (shield RL),  a shield synthesis based reinforcement learning algorithm, by automatically constructing safety envelopes from DC specifications. 


Correct-by-Construction Reinforcement Learning of Cardiac Pacemakers from Duration Calculus Requirements

In silico modeling of LAT1 efflux agonists for the treatment of pancreatic cancer

In our previous work, we identified clustering of kidney cancers based on differential metabolic pathway activation, and an association between metabolic clustering and survival in renal cell carcinoma. Here we further study clear cell renal carcinoma, papillary cell renal carcinoma and chromophobe renal cell carcinoma by applying machine learning models to pinpoint specific metabolic reactions, and their master regulators, that best explain differences in survival in these cancers. We constructed a cohort of 941 patients in The Cancer Genome Atlas (TCGA) diagnosed with kidney cancer (project codes KIPR [papillary cell RCC], KIRC [clear cell RCC] and KICH [chromophobe RCC]). We used tumor RNA-Seq data to construct reaction activity scores (RAS) using metabolic genes according to the HMRcore rules and taking into account enzymatic subunits and multiple different genes coding for proteins that catalyze the same reaction. We then used a machine learning model to estimate survival as a function of metabolic pathways, age and gender. By using Cox regularization with elastic-net penalty we were able to simultaneously perform variable selection and estimation of survival in presence of a large number of regressors. Using master regulatory analysis, we estimated activity of key transcription factors and then used a multinomial model with elastic net penalty to identify those key transcription factors that have the greatest impact on dysregulation of tumor metabolic pathways in renal cell carcinomas. Using glucose transport as a known hallmark of renal cell carcinoma, we identified 19 metabolic reactions that are better predictors of survival than glucose transport in our machine learning model. The metabolic reactions we identified fit into larger categories that include folate and protein metabolism, as well as beta-oxidation. These are consistent with metabolic changes induced by mTOR, and it is notable that inhibitors of mTOR are already used in the clinic for treatment of renal cell carcinoma. Our research raises the question of whether effectiveness of mTOR inhibitors can be increased by amplifying the effects of mTOR inhibition on oncogenesis through targeting of related master regulators of metabolism.

Analysis of Metabolic Profile and Survival Data in Renal Cancer

Background
TCEP is an organophosphorus flame-retardant (OPFR) widely used in polyurethane foams and household products. From these materials, TCEP emits into the surrounding environment, which humans encounter through inhalation and dermal absorption. Previous studies have shown OPFRs to have endocrine disrupting actions, and thus, influence diseases such as cancer. Breast cancer is the second most common cancer among women in the United States, and because of this prevalence, it is imperative to investigate possible carcinogens. Due to the endocrine disrupting nature of OPFRs, we are investigating the effects of TCEP on hormone-dependent breast cancer cell lines MCF-7 and T-47D.

Methods
Our study examines the effects of TCEP, alone and in combination with hormones and anti- hormones, on ERα and BRCA-1 expression in MCF-7 and T-47D breast cancer cell lines by utilizing western blot analyses, cellular viability assays, confocal microscopy, and RT-qPCR analyses. In order to deplete any endogenous steroids or effectors, breast cancer cells were cultured in a medium containing 5% charcoal-stripped fetal bovine serum for six days.

Results
Western blot analysis revealed alterations in ERα and BRCA-1 expression after 24 hours of treatment with varying TCEP concentrations (1μM-2mM). Analysis revealed a concentration- dependent increase in BRCA-1 expression in both cell lines, and a concentration-dependent decrease in ERα expression in the T-47D cell line. During hormonal studies, cells were treated with an optimum concentration of TCEP along with combinations of hormones and anti- hormones. After 24-hour treatment with a combination of E2 + TCEP, a significant decrease in ERα expression was observed in both cell lines. A significant increase of BRCA-1 expression occurred after 24-hour treatment of TCEP + E2 in both cell lines. These effects were sensitive to the presence of antiestrogens. Cellular viability assays revealed a significant increase in cellular viability in both cell lines when treated with a combination of E2 + TCEP.

Conclusion
Our results suggest that TECP acts as an endocrine disrupting compound on the steroid receptors and tumor suppressor genes in breast cancer cells. Further studies are warranted to determine the exact mechanism of action and influence of TCEP on breast cancer progression.

The Effects of Tris(2-Chloroethyl) phosphate (TCEP) on the Expression of Estrogen Receptor α (ERα) and Tumor Suppressor BRCA-1 in MCF-7 and T-47D Breast Cancer Cell Lines

Background
Glioblastoma (GBM) represents the most common primary malignant brain tumor in adults, with a median survival rate of less than two years. GBM presents with therapeutic resistance and high rates of recurrence, largely attributed to the presence of glioblastoma stem cells (GSCs). GSCs are frequently located in specific niches, including a hypoxic niche and a normoxic, perivascular niche. Targeting GSCs is a potential therapeutic avenue for treating GBM, but the details of GSC function in hypoxic conditions remains poorly understood. Further, most gene expression findings are limited to epigenetic or transcriptional regulation assays, overlooking the importance of the translatome. This study expands on conclusions made using steady-state mRNA levels by including ribosome profiling to evaluate translational regulation mechanisms. 

Methods
We selected two patient-derived GSC lines (GSC 215, GSC 316) and cultured them in both hypoxic (5% O2) and normoxic (21% O2) conditions. RNA-seq libraries were prepared in parallel with RIBO-seq libraries for the lines in both conditions, with biological replicates for each treatment group, after which Illumina sequencing was performed. Raw sequencing reads were assessed for quality. Single-end reads were aligned and counted using command line versions of STAR, BEDTools, and SAMTools. Differential changes in ribosome-protected fragments (∆RPF) and changes in transcript levels (∆RNA) were calculated using DESeq2. GO term analyses for transcripts was performed with Reactome using a hypergeometric statistical test and BH FDR correction.

Results
Comparing ∆RNA and ∆RPF levels between hypoxic and normoxic GSCs reveals 9 functional clusters of genes. These categories indicate genes which are post-transcriptionally controlled in 3 manners: translationally suppressed (28.19%), translationally upregulated (32.87%), or concurrently translated (38.93%). Altogether, more than 60% of mRNA show marked regulation on the translational level.

Significant enrichment of genes in the NOTCH signaling pathway (p = 1.09e-02), WNT signaling pathway (p = 1.37e-04), and RHO GTPase cycle (p = 2.37e-07) are present among the transcripts that experience translational upregulation in hypoxic conditions.

Conclusion
This study may lead to the identification of potential therapeutic targets as evidenced by their function as regulatory hubs in cellular networks responsible for hypoxic adaptation. These observations are largely undetected by RNA-seq measurements.


Evaluating the Role of Translation Regulation in Glioblastoma Stem Cells for Hypoxic Adaptation

Neuroblastoma is the most common extracranial solid tumor in children. It accounts for more than 7% of malignancies in patients less than fifteen years of age and makes up 15% of oncology deaths. Neuroblastoma is a cancer derived from primordial neural crest cells under conditions of inappropriate migration, maturation, and or differentiation. These extracranial tumors are formed in the adrenal medulla, sympathetic ganglia, and abdomen. Current treatments for neuroblastoma include surgery, chemotherapy, radiotherapy, and biotherapy. Interestingly, neuroblastoma tumors are heterogenous with some children having low-risk tumors that regress entirely while others have high-risk tumors with widespread metastasis and poor outcomes. The expression of MYCN, a gene in a family of oncogenic drivers, in neuroblastoma tumors has been associated with high-risk neuroblastoma and worse outcomes. Children with high-risk neuroblastoma account for approximately half of all patients diagnosed with neuroblastoma. Despite current interventions and aggressive treatment, children with high-risk neuroblastoma have a long-term survival rate of less than 50%. Currently, there are not many targeted therapies for high-risk neuroblastoma as there is little know about the mechanism driving this cancer. By identifying the mechanism driving tumor progression, a treatment to specifically target the different steps of this cancer can be created. Signaling through neuronal nicotinic acetylcholine receptors (nAChRs) is involved in many neuronal processes of neural tissue, including the proliferation of neural crest cells. These receptors are a diverse group of ligand-gated pentametric channels can exist as homopentamers or heteropentamers of α or β subunits. The only mammalian homopentamer is the α7 receptor that when activated is suggested to promote neural cell survival. Furthermore, expression of nAChR has been shown to be regulated by MYCN. Our lab is interested to see if MYCN-expressing neuroblastoma tumors upregulate the expression of nAChRs leading to an overall drive in tumor progression. Therefore, targeting this interaction between nAChRs and MYCN can become a potential therapy of high-risk neuroblastoma. 


N-MYC Transcription Factor Increases the Expression of Alpha-7 Nicotinic Acetylcholine Receptor Promoting Cell Viability in Neuroblastoma Cell Line

Background and Premise: Triple-negative breast cancer (TNBC) is an aggressive, metastatic disease with high mortality. The standard-of-care neoadjuvant chemotherapy (NACT) shows an initial response but followed by a drug resistance, increase in tumor relapse and distant metastases. Immunotherapy against programmed death ligand-1 (PD-L1) has shown efficacy in treating a subset of TNBC patients. However, it shows resistance in other patients due to new immune checkpoints such as VISTA, TIGIT, and LAG3. 

eIF4A1 is a mRNA helicase known to translate many oncogenic mRNAs including survivin, c-MYC, ROCK1, STAT1, ARF6 etc., We previously established that eIF4A1 is a vulnerable target in drug-resistant TNBC. Interestingly, we observed a reduction in levels of PD-L1 when eIF4A1 is targeted. Many downstream effectors of eIF4A1 such as c-MYC, STAT1 and ARF6 are indeed upstream regulators of PD-L1. This premise formed the basis to co-target eIF4A1 and PD-L1. Before that, we need to establish the clinical relevance of these targets in human TNBC biospecimens. This will ascertain the differential operation of the eIF4A1-PD-L1 axis in therapy-naïve and drug-resistant primary and mTNBC.

Approach: Human biospecimens obtained from CHTN NIH repositories (more than 100 TNBC primary tumor and metastatic biospecimens) will be analyzed for role of eIF4A1 in translation of oncogenic mRNAs that contribute to PD-L1 and other emerging immune checkpoint markers such as VISTA, TIGIT, and LAG3 via immunohistochemistry (after deglycosylation for effective detection of PD-L1) and also by immunoblotting. 

Results: The levels of total eIF4A1 from 16 matched, therapy naive TNBC tissue were compared to adjacent normal breast tissue from the same patient via immunoblotting. Total eIF4A1 levels were found to be significantly elevated in TNBC compared to adjacent normal tissue. The downstream effectors of eIF4A1 such as STAT1 and ARF6 were also significantly elevated indicating a robust helicase activity of eIF4A1. Importantly, PD-L1 levels was found to be elevated. Additionally, the total eIF4A1 level were elevated in drug-resistant tumors than normal breast tissue samples as assesses by immunohistochemistry. 

Conclusion: The eIF4A1-STAT1-PD-L1 axis is highly active in TNBC especially in drug-resistant situations. Currently, we are evaluating other immune checkpoints such as VISTA, TIGIT, and LAG3. 


Underline digital video library

LIVE - Cultural drivers of stone-tool use in Balinese long-tailed macaques

AniEE: A Dataset of Animal Experimental Literature for Event Extraction

User-Oriented Robust Reinforcement Learning

Evaluating Model-free Reinforcement Learning toward Safety-critical Tasks

Certified Policy Smoothing for Cooperative Multi-Agent Reinforcement Learning

Correct-by-Construction Reinforcement Learning of Cardiac Pacemakers from Duration Calculus Requirements

In silico modeling of LAT1 efflux agonists for the treatment of pancreatic cancer

Analysis of Metabolic Profile and Survival Data in Renal Cancer

The Effects of Tris(2-Chloroethyl) phosphate (TCEP) on the Expression of Estrogen Receptor α (ERα) and Tumor Suppressor BRCA-1 in MCF-7 and T-47D Breast Cancer Cell Lines

Evaluating the Role of Translation Regulation in Glioblastoma Stem Cells for Hypoxic Adaptation

N-MYC Transcription Factor Increases the Expression of Alpha-7 Nicotinic Acetylcholine Receptor Promoting Cell Viability in Neuroblastoma Cell Line

Operation of eIF4A1-PD-L1 axis in therapy-naïve and drug-resistant TNBC

Stay up to date with the latest Underline news!

PRESENTATIONS

CONFERENCES

COMPANY

RESOURCES