arXiv subjects

Fabrizio Costa

Publications and source records attributed to Fabrizio Costa.

4 recordsLinked to original sources

arXiv2025

Learning to generate feasible graphs using graph grammars

Generative methods for graphs need to be sufficiently flexible to model complex dependencies between sets of nodes. At the same time, the generated graphs need to satisfy domain-dependent feasibility conditions, that is, they should not violate certain constraints that would make their interpretation impossible within the given application domain (e.g. a molecular graph where an atom has a very large number of chemical bounds). Crucially, constraints can involve not only local but also long-range dependencies: for example, the maximal length of a cycle can be bounded. Currently, a large class of generative approaches for graphs, such as methods based on artificial neural networks, is based on message passing schemes. These approaches suffer from information 'dilution' issues that severely limit the maximal range of the dependencies that can be modeled. To address this problem, we propose a generative approach based on the notion of graph grammars. The key novel idea is to introduce a domain-dependent coarsening procedure to provide short-cuts for long-range dependencies. We show the effectiveness of our proposal in two domains: 1) small drugs and 2) RNA secondary structures. In the first case, we compare the quality of the generated molecular graphs via the Molecular Sets (MOSES) benchmark suite, which evaluates the distance between generated and real molecules, their lipophilicity, synthesizability, and drug-likeness. In the second case, we show that the approach can generate very large graphs (with hundreds of nodes) that are accepted as valid examples for a desired RNA family by the "Infernal" covariance model, a state-of-the-art RNA classifier. Our implementation is available on github: github.com/fabriziocosta/GraphLearn

cs.LG↗

arXiv2024

ProtScan: Modeling and Prediction of RNA-Protein Interactions

CLIP-seq methods are valuable techniques to experimentally determine transcriptome-wide binding sites of RNA-binding proteins. Despite the constant improvement of such techniques (e.g. eCLIP), the results are affected by various types of noise and depend on experimental conditions such as cell line, tissue, gene expression levels, stress conditions etc., paving the way for the in silico modeling of RNA-protein interactions. Here we present ProtScan, a predictive tool based on consensus kernelized SGD regression. ProtScan denoises and generalizes the information contained in CLIP-seq experiments. It outperforms competitor state-of the-art methods and can be used to model RNA-protein interactions on a transcriptome-wide scale.

q-bio.BM↗

arXiv2016

AptaTRACE: Elucidating Sequence-Structure Binding Motifs by Uncovering Selection Trends in HT-SELEX Experiments

Aptamers, short synthetic RNA/DNA molecules binding specific targets with high affinity and specificity, are utilized in an increasing spectrum of bio-medical applications. Aptamers are identified in vitro via the Systematic Evolution of Ligands by Exponential Enrichment (SELEX) protocol. SELEX selects binders through an iterative process that, starting from a pool of random ssDNA/RNA sequences, amplifies target-affine species through a series of selection cycles. HT-SELEX, which combines SELEX with high throughput sequencing, has recently transformed aptamer development and has opened the field to even more applications. HT-SELEX is capable of generating over half a billion data points, challenging computational scientists with the task of identifying aptamer properties such as sequence structure motifs that determine binding. While currently available motif finding approaches suggest partial solutions to this question, none possess the generality or scalability required for HT-SELEX data, and they do not take advantage of important properties of the experimental procedure. We present AptaTRACE, a novel approach for the identification of sequence-structure binding motifs in HT-SELEX derived aptamers. Our approach leverages the experimental design of the SELEX protocol and identifies sequence-structure motifs that show a signature of selection. Because of its unique approach, AptaTRACE can uncover motifs even when these are present in only a minuscule fraction of the pool. Due to these features, our method can help to reduce the number of selection cycles required to produce aptamers with the desired properties, thus reducing cost and time of this rather expensive procedure. The performance of the method on simulated and real data indicates that AptaTRACE can detect sequence-structure motifs even in highly challenging data.

q-bio.QM↗

arXiv2014

kLog: A Language for Logical and Relational Learning with Kernels

We introduce kLog, a novel approach to statistical relational learning. Unlike standard approaches, kLog does not represent a probability distribution directly. It is rather a language to perform kernel-based learning on expressive logical and relational representations. kLog allows users to specify learning problems declaratively. It builds on simple but powerful concepts: learning from interpretations, entity/relationship data modeling, logic programming, and deductive databases. Access by the kernel to the rich representation is mediated by a technique we call graphicalization: the relational representation is first transformed into a graph --- in particular, a grounded entity/relationship diagram. Subsequently, a choice of graph kernel defines the feature space. kLog supports mixed numerical and symbolic data, as well as background knowledge in the form of Prolog or Datalog programs as in inductive logic programming systems. The kLog framework can be applied to tackle the same range of tasks that has made statistical relational learning so popular, including classification, regression, multitask learning, and collective classification. We also report about empirical comparisons, showing that kLog can be either more accurate, or much faster at the same level of accuracy, than Tilde and Alchemy. kLog is GPLv3 licensed and is available at http://klog.dinfo.unifi.it along with tutorials.

cs.AI↗