Functional Inference Research Articles

Whole-blood transcriptome unveils altered immune response in acute myocardial infarction patients with aortic valve sclerosis

Abstract Funding Acknowledgements Type of funding sources: Foundation. Main funding source(s): Fondazione Umberto Veronesi (Research Grant 2011-12 to GIC) Fondazione Gigi e Pupa Ferrari ONLUS (FPF-14 to PP) Background Aortic valve sclerosis (AVSc) presents similar pathogenetic mechanisms to coronary artery disease (CAD) and is associated with short- and long-term mortality in CAD patients. Evidence of AVSc-specific pathophysiological traits in acute myocardial infarction (AMI) is currently lacking. Purpose We aimed to identify a blood-based transcriptional signature that could differentiate AVSc from non-AVSc patients during AMI. Methods Whole-blood transcriptome of AVSc (n = 44) and no-AVSc (n = 66) patients with AMI was assessed by RNA-sequencing on hospital admission. Feature selection, differential expression, and enrichment analyses were performed to identify gene expression patterns discriminating AVSc from no-AVSc and infer functional associations. Multivariable Cox regression analysis was used to estimate the hazard ratios of cardiovascular events in AVSc versus no-AVSc patients. Results This cross-sectional study identified a panel of 100 informative genes capable of distinguishing AVSc from no-AVSc patients with 94% accuracy. Further analysis revealed significant mean differences in 143 genes, of which 30 genes withstood correction for age and previous AMI or coronary interventions. Functional inference unveiled a significant association between AVSc and key biological processes, including acute inflammatory responses, type I interferons (IFN) response, platelet activation, and hemostasis. Notably, AMI patients with AVSc exhibited a significantly higher incidence of adverse cardiovascular events during a 10-year follow-up period, with a full adjusted hazard ratio of 2.4 (95% CI: 1.3–4.5). Conclusions Our findings shed light on the molecular mechanisms underlying AVSc and provide potential prognostic insights for AMI patients with AVSc. During AMI, AVSc patients showed increased type I IFN response and earlier adverse cardiovascular outcomes. Novel pharmacological therapies aiming at limiting type I IFN response during or immediately after AMI might improve poor cardiovascular outcomes of AMI patients with AVSc.

Zero-shot urban function inference with street view images through prompting a pretrained vision-language model

Inferring urban functions using street view images (SVIs) has gained tremendous momentum. The recent prosperity of large-scale vision-language pretrained models sheds light on addressing some long-standing challenges in this regard, for example, heavy reliance on labeled samples and computing resources. In this paper, we present a novel prompting framework for enabling the pretrained vision-language model CLIP to effectively infer fine-grained urban functions with SVIs in a zero-shot manner, that is, without labeled samples and model training. The prompting framework UrbanCLIP comprises an urban taxonomy and several urban function prompt templates, in order to (1) bridge the abstract urban function categories and concrete urban object types that can be readily understood by CLIP, and (2) mitigate the interference in SVIs, for example, street-side trees and vehicles. We conduct extensive experiments to verify the effectiveness of UrbanCLIP. The results indicate that the zero-shot UrbanCLIP largely surpasses several competitive supervised baselines, e.g. a fine-tuned ResNet, and its advantages become more prominent in cross-city transfer tests. In addition, UrbanCLIP’s zero-shot performance is considerably better than the vanilla CLIP. Overall, UrbanCLIP is a simple yet effective framework for urban function inference, and showcases the potential of foundation models for geospatial applications.

Holed up, but thriving: Impact of multitrophic cryoconite communities on glacier elemental cycles

Cryoconite holes (water and sediment-filled depressions), found on glacier surfaces worldwide, serve as reservoirs of microbes, carbon, trace elements, and nutrients, transferring these components downstream via glacier hydrological networks. Through targeted amplicon sequencing of carbon and nitrogen cycling genes, coupled with functional inference-based methods, we explore the functional diversity of these mini-ecosystems within Antarctica and the Himalayas. These regions showcase distinct environmental gradients and experience varying rates of environmental change influenced by global climatic shifts. Analysis revealed a diverse array of photosynthetic microorganisms, including Stramenopiles, Cyanobacteria, Rhizobiales, Burkholderiales, and photosynthetic purple sulfur Proteobacteria. Functional inference highlighted the high potential for carbohydrate, amino acid, and lipid metabolism in the Himalayan region, where organic carbon concentrations surpassed those in Antarctica by up to 2 orders of magnitude. Nitrogen cycling processes, including fixation, nitrification, and denitrification, are evident, with Antarctic cryoconite exhibiting a pronounced capacity for nitrogen fixation, potentially compensating for the limited nitrate concentrations in this region. Processes associated with the respiration of elemental sulfur and inorganic sulfur compounds such as sulfate, sulfite, thiosulfate, and sulfide suggest the presence of a complete sulfur cycle. The Himalayan region exhibits a higher potential for sulfur cycling, likely due to the abundant sulfate ions and sulfur-bearing minerals in this region. The capability for complete iron cycling through iron oxidation and reduction reactions was also predicted. Methanogenic archaea that produce methane during organic matter decomposition and methanotrophic bacteria that utilize methane as carbon and energy sources co-exist in the cryoconite, suggesting that these niches support the complete cycling of methane. Additionally, the presence of various microfauna suggests the existence of a complex food web. Collectively, these results indicate that cryoconite holes are self-sustaining ecosystems that drive elemental cycles on glaciers and potentially control carbon, nitrogen, sulfur, and iron exports downstream.

Improvements in viral gene annotation using large language models and soft alignments

BackgroundThe annotation of protein sequences in public databases has long posed a challenge in molecular biology. This issue is particularly acute for viral proteins, which demonstrate limited homology to known proteins when using alignment, k-mer, or profile-based homology search approaches. A novel methodology employing Large Language Models (LLMs) addresses this methodological challenge by annotating protein sequences based on embeddings.ResultsCentral to our contribution is the soft alignment algorithm, drawing from traditional protein alignment but leveraging embedding similarity at the amino acid level to bypass the need for conventional scoring matrices. This method not only surpasses pooled embedding-based models in efficiency but also in interpretability, enabling users to easily trace homologous amino acids and delve deeper into the alignments. Far from being a black box, our approach provides transparent, BLAST-like alignment visualizations, combining traditional biological research with AI advancements to elevate protein annotation through embedding-based analysis while ensuring interpretability. Tests using the Virus Orthologous Groups and ViralZone protein databases indicated that the novel soft alignment approach recognized and annotated sequences that both blastp and pooling-based methods, which are commonly used for sequence annotation, failed to detect.ConclusionThe embeddings approach shows the great potential of LLMs for enhancing protein sequence annotation, especially in viral genomics. These findings present a promising avenue for more efficient and accurate protein function inference in molecular biology.

A MCMC method based on surrogate model and Gaussian process parameterization for infinite Bayesian PDE inversion

This work focuses on an inversion problem derived from parametric partial differential equations (PDEs) with an infinite-dimensional parameter, represented as a coefficient function. The objective is to estimate this coefficient function accurately despite having only noisy measurements of the PDE solution at sparse input points. Conventional methods for inversion require numerous calls to a refined PDE solver, resulting in significant computational complexity, especially for challenging PDE problems, making the inference of the coefficient function practically infeasible. To address this issue, we propose a MCMC method that combines an deep learning-based surrogate and Gaussian process parameterization to efficiently infer the posterior of the coefficient function. The surrogate model is a combination of a cost-effective coarse PDE solver and a neural network-based transformation which provides an approximate solution derived from the refined PDE solver but based on the coarse PDE solution. The coefficient function is represented by Gaussian process with finite number of spatially dependent parameters and this parametric representation will be beneficial for the preconditioned Crank-Nicolson (pCN) Markov chain Monte Carlo method to sample efficiently from the posterior distribution. Approximate Bayesian computation method is used for constructing an informative dataset for the transformation learning. Our numerical examples demonstrate the effectiveness of this approach in accurately estimating the coefficient function while significantly reducing the computational burden associated with traditional inversion techniques.

Assembly collapsing versus heterozygosity oversizing: detection of homokaryotic and heterokaryotic Laccaria trichodermophora strains by hybrid genome assembly.

Genome assembly and annotation using short-paired reads is challenging for eukaryotic organisms due to their large size, variable ploidy and large number of repetitive elements. However, the use of single-molecule long reads improves assembly quality (completeness and contiguity), but haplotype duplications still pose assembly challenges. To address the effect of read length on genome assembly quality, gene prediction and annotation, we compared genome assemblers and sequencing technologies with four strains of the ectomycorrhizal fungus Laccaria trichodermophora. By analysing the predicted repertoire of carbohydrate enzymes, we investigated the effects of assembly quality on functional inferences. Libraries were generated using three different sequencing platforms (Illumina Next-Seq, Mi-Seq and PacBio Sequel), and genomes were assembled using single and hybrid assemblies/libraries. Long reads or hybrid assemby resolved the collapsing of repeated regions, but the nuclear heterozygous versions remained unresolved. In dikaryotic fungi, each cell includes two nuclei and each nucleus has differences not only in allelic gene version but also in gene composition and synteny. These heterokaryotic cells produce fragmentation and size overestimation of the genome assembly of each nucleus. Hybrid assembly revealed a wider functional diversity of genomes. Here, several predicted oxidizing activities on glycosyl residues of oligosaccharides and several chitooligosaccharide acetylase activities would have passed unnoticed in short-read assemblies. Also, the size and fragmentation of the genome assembly, in combination with heterozygosity analysis, allowed us to distinguish homokaryotic and heterokaryotic strains isolated from L. trichodermophora fruit bodies.

Deciphering the Structural and Functional Diversity of Rhizobacteria from Stone Pine Inoculated with Plant Growth Promoting Rhizobacteria (PGPR) before and after Transplanted into Degraded Agricultural Soil

The use of plant growth-promoting rhizobacteria (PGPR) inoculated on plants has shown that it can increase the success of reforestation and accelerate soil recovery by improving soil microbial diversity. Three PGPR isolated from natural pine populations were selected for their metabolic capabilities and taxonomic affiliation (Z4.3; Bacillus sp., Z5.4; Arthobacter sp., and Z7.15; and Pseudomonas sp.) when inoculated alone or in combination (consortium) on stone pine seedlings before transplanting to the field. Before transplanting and after nine months, rhizospheric soil samples were collected for structural and functional metagenomic studies. First, the data were analyzed using EasyMAP. Neither alpha nor beta diversity showed significant differences between the samples, although unique taxa representative of each sample were detected. The predominant phylum in all cases was Proteobacteria, followed by Bacteroidetes and Acidobacteria. The linear discriminant analysis (LDA) effect size (LEfSe) found significantly over-represented taxa in some samples, highlighting different representatives of the order Sphingomonadales in several of them. Functional inference performed with PICRUSt also showed significantly over-represented functions in some samples. The study demonstrates that PGPR have a positive effect on plants and cause detectable changes in microbial communities in terms of both structure and function.

Functional genomic screens with death rate analyses reveal mechanisms of drug action.

A common approach for understanding how drugs induce their therapeutic effects is to identify the genetic determinants of drug sensitivity. Because 'chemo-genetic profiles' are performed in a pooled format, inference of gene function is subject to several confounding influences related to variation in growth rates between clones. In this study, we developed Method for Evaluating Death Using a Simulation-assisted Approach (MEDUSA), which uses time-resolved measurements, along with model-driven constraints, to reveal the combination of growth and death rates that generated the observed drug response. MEDUSA is uniquely effective at identifying death regulatory genes. We apply MEDUSA to characterize DNA damage-induced lethality in the presence and absence of p53. Loss of p53 switches the mechanism of DNA damage-induced death from apoptosis to a non-apoptotic death that requires high respiration. These findings demonstrate the utility of MEDUSA both for determining the genetic dependencies of lethality and for revealing opportunities to potentiate chemo-efficacy in a cancer-specific manner.

Large-scale correlation screening under dependence for brain functional connectivity network inference

Large-scale correlation screening under dependence for brain functional connectivity network inference

MARS and RNAcmap3: The Master Database of All Possible RNA Sequences Integrated with RNAcmap for RNA Homology Search.

Recent success of AlphaFold2 in protein structure prediction relied heavily on co-evolutionary information derived from homologous protein sequences found in the huge, integrated database of protein sequences (Big Fantastic Database). In contrast, the existing nucleotide databases were not consolidated to facilitate wider and deeper homology search. Here, we built a comprehensive database by incorporating the non-coding RNA (ncRNA) sequences from RNAcentral, the transcriptome assembly and metagenome assembly from metagenomics RAST (MG-RAST), the genomic sequences from Genome Warehouse (GWH), and the genomic sequences from MGnify, in addition to the nucleotide (nt) database and its subsets in National Center of Biotechnology Information (NCBI). The resulting Master database of All possible RNA sequences (MARS) is 20-fold larger than NCBI's nt database or 60-fold larger than RNAcentral. The new dataset along with a new split-search strategy allows a substantial improvement in homology search over existing state-of-the-art techniques. It also yields more accurate and more sensitive multiple sequence alignments (MSAs) than manually curated MSAs from Rfam for the majority of structured RNAs mapped to Rfam. The results indicate that MARS coupled with the fully automatic homology search tool RNAcmap will be useful for improved structural and functional inference of ncRNAs and RNA language models based on MSAs. MARS is accessible at https://ngdc.cncb.ac.cn/omix/release/OMIX003037, and RNAcmap3 is accessible at http://zhouyq-lab.szbl.ac.cn/download/.

On the limits of 16S rRNA gene-based metagenome prediction and functional profiling.

Molecular profiling techniques such as metagenomics, metatranscriptomics or metabolomics offer important insights into the functional diversity of the microbiome. In contrast, 16S rRNA gene sequencing, a widespread and cost-effective technique to measure microbial diversity, only allows for indirect estimation of microbial function. To mitigate this, tools such as PICRUSt2, Tax4Fun2, PanFP and MetGEM infer functional profiles from 16S rRNA gene sequencing data using different algorithms. Prior studies have cast doubts on the quality of these predictions, motivating us to systematically evaluate these tools using matched 16S rRNA gene sequencing, metagenomic datasets, and simulated data. Our contribution is threefold: (i) using simulated data, we investigate if technical biases could explain the discordance between inferred and expected results; (ii) considering human cohorts for type two diabetes, colorectal cancer and obesity, we test if health-related differential abundance measures of functional categories are concordant between 16S rRNA gene-inferred and metagenome-derived profiles and; (iii) since 16S rRNA gene copy number is an important confounder in functional profiles inference, we investigate if a customised copy number normalisation with the rrnDB database could improve the results. Our results show that 16S rRNA gene-based functional inference tools generally do not have the necessary sensitivity to delineate health-related functional changes in the microbiome and should thus be used with care. Furthermore, we outline important differences in the individual tools tested and offer recommendations for tool selection.

Curation, nomenclature, and topological classification of receptor-like kinases from 528 plant species for novel domain discovery and functional inference

Receptor-like kinases (RLKs) are the most numerous signal transduction components in plants and play important roles in determining how different plants adapt to their ecological environments. Research on RLKs has focused mainly on a small number of typical RLK members in a few model plants. There is an urgent need to study the composition, distribution, and evolution of RLKs at the holistic level to increase our understanding of how RLKs assist in the ecological adaptations of different plant species. In this study, we collected the genome assemblies of 528 plant species and constructed an RLK dataset. Using this dataset, we identified and characterized 524 948 RLK family members. Each member underwent systematic topological classification and was assigned a gene ID based on a unified nomenclature system. Furthermore, we identified two novel extracellular domains in some RLKs, designated Xiao and Xiang. Evolutionary analysis of the RLK family revealed that the RLCK-XVII and RLCK-XII-2 classes were present exclusively in dicots, suggesting that diversification of RLKs between monocots and dicots may have led to differences in downstream cytoplasmic responses. We also used an interaction proteome to help empower data mining for inference of new RLK functions from a global perspective, with the ultimate goal of understanding how RLKs shape the adaptation of different plants to the environments/ecosystems. The assembled RLK dataset, together with annotations and analytical tools, forms an integrated foundation of multiomics data that is publicly accessible via the metaRLK web portal (http://metaRLK.biocloud.top).

A landscape of gene expression regulation for synovium in arthritis

The synovium is an important component of any synovial joint and is the major target tissue of inflammatory arthritis. However, the multi-omics landscape of synovium required for functional inference is absent from large-scale resources. Here we integrate genomics with transcriptomics and chromatin accessibility features of human synovium in up to 245 arthritic patients, to characterize the landscape of genetic regulation on gene expression and the regulatory mechanisms mediating arthritic diseases predisposition. We identify 4765 independent primary and 616 secondary cis-expression quantitative trait loci (cis-eQTLs) in the synovium and find that the eQTLs with multiple independent signals have stronger effects and heritability than single independent eQTLs. Integration of genome-wide association studies (GWASs) and eQTLs identifies 84 arthritis related genes, revealing 38 novel genes which have not been reported by previous studies using eQTL data from the GTEx project or immune cells. We further develop a method called eQTac to identify variants that could affect gene expression by affecting chromatin accessibility and identify 1517 regions with potential regulatory function of chromatin accessibility. Altogether, our study provides a comprehensive synovium multi-omics resource for arthritic diseases and gains new insights into the regulation of gene expression.

Model-Based Smoothing with Integrated Wiener Processes and Overlapping Splines

In many applications that involve the inference of an unknown smooth function, the inference of its derivatives is also important. To make joint inferences of the function and its derivatives, a class of Gaussian processes called pth order Integrated Wiener’s Process (IWP), is considered. Methods for constructing a finite element (FEM) approximation of an IWP exist but only focus on the case p = 2 and do not allow appropriate inference for derivatives. In this article, we propose an alternative FEM approximation with overlapping splines (O-spline). The O-spline approximation applies for any order p ∈ Z + , and provides consistent and efficient inference for all derivatives up to order p – 1. It is shown both theoretically and empirically that the O-spline approximation converges to the IWP as the number of knots increases. We further provide a unified and interpretable way to define priors for the smoothing parameter based on the notion of predictive standard deviation, which is invariant to the order p and the knot placement. Finally, we demonstrate the practical use of the O-spline approximation through an analysis of COVID death rates where the inference of derivative has an important interpretation in terms of the course of the pandemic. Supplementary materials for this article are available online.

AlignScape, displaying sequence similarity using self-organizing maps.

The current richness of sequence data needs efficient methodologies to display and analyze the complexity of the information in a compact and readable manner. Traditionally, phylogenetic trees and sequence similarity networks have been used to display and analyze sequences of protein families. These methods aim to shed light on key computational biology problems such as sequence classification and functional inference. Here, we present a new methodology, AlignScape, based on self-organizing maps. AlignScape is applied to three large families of proteins: the kinases and GPCRs from human, and bacterial T6SS proteins. AlignScape provides a map of the similarity landscape and a tree representation of multiple sequence alignments These representations are useful to display, cluster, and classify sequences as well as identify functional trends. The efficient GPU implementation of AlignScape allows the analysis of large MSAs in a few minutes. Furthermore, we show how the AlignScape analysis of proteins belonging to the T6SS complex can be used to predict coevolving partners.

Computational inference of eIF4F complex function and structure in human cancers

The canonical eukaryotic initiation factor 4F (eIF4F) complex, composed of eIF4G1, eIF4A1, and the cap-binding protein eIF4E, plays a crucial role in cap-dependent translation initiation in eukaryotic cells. An alternative cap-independent initiation can occur, involving only eIF4G1 and eIF4A1 through internal ribosome entry sites (IRESs). This mechanism is considered complementary to cap-dependent initiation, particularly in tumors under stress conditions. However, the selection and molecular mechanism of specific translation initiation remains poorly understood in human cancers. Thus, we analyzed gene copy number variations (CNVs) in TCGA tumor samples and found frequent amplification of genes involved in translation initiation. Copy number gains in EIF4G1 and EIF3E frequently co-occur across human cancers. Additionally, EIF4G1 expression strongly correlates with genes from cancer cell survival pathways including cell cycle and lipogenesis, in tumors with EIF4G1 amplification or duplication. Furthermore, we revealed that eIF4G1 and eIF4A1 protein levels strongly co-regulate with ribosomal subunits, eIF2, and eIF3 complexes, while eIF4E co-regulates with 4E-BP1, ubiquitination, and ESCRT proteins. Utilizing Alphafold predictions, we modeled the eIF4F structure with and without eIF4E binding. For cap-dependent initiation, our modeling reveals extensive interactions between the N-terminal eIF4E-binding domain of eIF4G1 and eIF4E. Furthermore, the eIF4G1 HEAT-2 domain positions eIF4E near the eIF4A1 N-terminal domain (NTD), resulting in the collaborative enclosure of the RNA binding cavity within eIF4A1. In contrast, during cap-independent initiation, the HEAT-2 domain directly binds the eIF4A1-NTD, leading to a stronger interaction between eIF4G1 and eIF4A1, thus closing the mRNA binding cavity without the involvement of eIF4E.

Utilising energy function and variational inference training for learning a graph neural network architecture

Utilising energy function and variational inference training for learning a graph neural network architecture

Play Behavior Varies with Age, Sex, and Socioecological Context in Wild, Immature Orangutans (Pongo spp.)

Play is thought to serve different purposes at different times during ontogeny. The nature and frequency of play are expected to change accordingly over the developmental trajectory and with socio-ecological context. Orangutans offer the opportunity to disentangle the ontogenetic trajectories of solitary and social play with their extended immature phase, and socio-ecological variation among populations and species. We evaluated the frequency of play in 39 immature individuals across two populations (Pongo pygmaeus wurmbii, at Tuanan, Borneo, and P. abelii at Suaq, Sumatra), age (0–11 years), sex, and social context, using more than 11 500 h of full-day focal observation data. We found independent age trajectories of different play types, with solitary object and solitary locomotor peaking before social play. Social play partners changed during ontogeny, and male immatures were more likely to engage in non-mother social play than females. Overall, social play was more frequent at Suaq than Tuanan, linked to the more frequent availability of partners. Furthermore, per time in association with conspecifics, Tuanan immatures were as likely to engage in social play as their peers at Suaq, suggesting similar intrinsic motivation. Increasing fruit availability correlated with both longer associations and increased social play frequency in the less sociable population of Tuanan, but not at Suaq. Our findings on orangutans support evidence from other species that different play types follow different developmental trajectories, vary with sex, social opportunities, and ecological context. Although drawing functional inferences is challenging, the distinct developmental trajectories reflecting adult sociability and behavioral repertoires may indicate that play serves several, non-mutually exclusive functions during ontogeny.

Kernel-, mean-, and noise-marginalized Gaussian processes for exoplanet transits and H0 inference

ABSTRACT Using a fully Bayesian approach, Gaussian process regression is extended to include marginalization over the kernel choice and hyperparameters. In addition, Bayesian model comparison via the evidence enables direct kernel comparison. The calculation of the joint posterior was implemented with a transdimensional sampler which simultaneously samples over the discrete kernel choice and their hyperparameters by embedding these in a higher dimensional space, from which samples are taken using nested sampling. Kernel recovery and mean function inference were explored on synthetic data from exoplanet transit light-curve simulations. Subsequently, the method was extended to marginalization over mean functions and noise models and applied to the inference of the present-day Hubble parameter, H0, from real measurements of the Hubble parameter as a function of redshift, derived from the cosmologically model-independent cosmic chronometer and lambda-cold dark matter-dependent baryon acoustic oscillation observations. The inferred H0 values from the cosmic chronometers, baryon acoustic oscillations, and combined data sets are $H_0= 66 \pm 6,\, 67 \pm 10,\, \mathrm{ and}\,69 \pm 6\,\mathrm{km}\, \mathrm{s}^{-1}\, \mathrm{Mpc}^{-1}$, respectively. The kernel posterior of the cosmic chronometers data set prefers a non-stationary linear kernel. Finally, the data sets are shown to be not in tension with ln R = 12.17 ± 0.02.

Protein Sequence Classification Through Deep Learning and Encoding Strategies

Protein sequence classification is vital for understanding protein functionalities, aiding in the inference of novel protein functions. Machine learning and deep learning algorithms have revolutionized this field, offering insights into specific protein classes and functions. This study employs Natural Language Processing (NLP) techniques, including Integer and Blosum encoding, for efficient classification. SVM with count vectorizer achieves the highest accuracy of 92%, while Integer encoding with CNN surpasses NLP embedding techniques by 4%. The goal is to develop an automated system for predicting protein functionality based on sequence classification, contributing to advancements in proteomics and computational biology.