Diogo de Jesus Soares Machado
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Research Subjects
Diogo de Jesus Soares Machado has a total of 12 co-authors. Here are the top co-authors by number of shared publications:
- Antonio Camilo da Silva Filho
Department of Clinical Analysis, Federal University of Paraná, Curitiba, Brazil
2 shared publications - Camilla Reginatto De Pierri
Affiliations Laboratory of Artificial Intelligence Applied to Bioinformatics, SEPT, Federal University of Paraná, Curitiba, Paraná, Brazil, Department of Biochemistry and Molecular Biology, Federal University of Paraná, Curitiba, Paraná, Brazil
2 shared publications - Roberto Tadeu Raittz
Laboratory of Artificial Intelligence Applied to Bioinformatics, Professional and Technological Education Sector (SEPT), Federal University of Paraná, Curitiba, Brazil
2 shared publications - Camila Pereira Perico
Graduate Program in Bioinformatics, SEPT, Federal University of Paraná, Curitiba, Paraná, Brazil; Laboratory of Artificial Intelligence Applied to Bioinformatics, SEPT, Federal University of Paraná, Curitiba, Paraná, Brazil
1 shared publication - Cyntia Maria Telles Fadel-Picheth
Departamento de Patologia Médica, Universidade Federal do Paraná, Curitiba, PR, Brasil
1 shared publication - Fabio Oliveira Pedrosa
Universidade Federal do Paraná
1 shared publication
Latest Publications
- Open Access
- 20 January 2026
- Published by Sociedade Brasileira de Computação in Journal of the Brazilian Computer Society
- Vol. 32 (1) , 1155-1164
Biotext with SWeePtex: Bioinformatics Tricks to Perform Fast, Accurate, and Content-specific String Embedding
The escalating demand for adaptable Artificial Intelligence (AI) systems presents a critical hurdle: generating efficient text embeddings tailored to specific problems. While Large Language Models (LLMs) excel in general contexts, they struggle in specialized domains due to their massive data requirements, opaque embedding strategies, and high computational costs. We introduce Biotext, featuring SWeePtex, a novel framework that adapts successful Bioinformatics techniques for text embedding. By converting text to the Biological Sequence-Like (BSL) format, our Python package enables the application of SWeeP, a tool originally developed for biological sequences, to create content-addressable vectors in natural language, employing the random projection paradigm. Using unsupervised machine learning, we validated this finding by analyzing data from 14,984 MEDLINE abstracts on the thioredoxin theme. Biotext, through SWeePtex, constructs a unified vector space for words and documents from scratch, capturing rich contextual relationships and offering scalable processing. Our usage example demonstrates that this Bioinformatics-inspired method effectively addresses key challenges in Natural Language Processing (NLP), providing interpretable, computationally efficient, and content-addressable linguistic representations for document exploration. Ultimately, Biotext demonstrates that bridging Bioinformatics and NLP yields powerful, efficient, and accessible text analysis tools that balance analytical power with interpretability, particularly valuable in specialized domains and resource-constrained environments. Biotext Python package is freely available at the PyPI repository.
- 20 January 2026
- Published by Sociedade Brasileira de Computação in Journal of the Brazilian Computer Society
- Vol. 32 (1) , 1155-1164
Open AccessBiotext with SWeePtex: Bioinformatics Tricks to Perform Fast, Accurate, and Content-specific String Embedding
The escalating demand for adaptable Artificial Intelligence (AI) systems presents a critical hurdle: generating efficient text embeddings tailored to specific problems. While Large Language Models (LLMs) excel in general contexts, they struggle in specialized domains due to their massive data requirements, opaque embedding strategies, and high computational costs. We introduce Biotext, featuring SWeePtex, a novel framework that adapts successful Bioinformatics techniques for text embedding. By converting text to the Biological Sequence-Like (BSL) format, our Python package enables the application of SWeeP, a tool originally developed for biological sequences, to create content-addressable vectors in natural language, employing the random projection paradigm. Using unsupervised machine learning, we validated this finding by analyzing data from 14,984 MEDLINE abstracts on the thioredoxin theme. Biotext, through SWeePtex, constructs a unified vector space for words and documents from scratch, capturing rich contextual relationships and offering scalable processing. Our usage example demonstrates that this Bioinformatics-inspired method effectively addresses key challenges in Natural Language Processing (NLP), providing interpretable, computationally efficient, and content-addressable linguistic representations for document exploration. Ultimately, Biotext demonstrates that bridging Bioinformatics and NLP yields powerful, efficient, and accessible text analysis tools that balance analytical power with interpretability, particularly valuable in specialized domains and resource-constrained environments. Biotext Python package is freely available at the PyPI repository.
- Open Access
- 29 November 2021
- Published by Frontiers Media SA in Frontiers in Microbiology
- Vol. 12
Prediction and Analysis in silico of Genomic Islands in Aeromonas hydrophila
Aeromonas are Gram-negative rods widely distributed in the environment. They can cause severe infections in fish related to financial losses in the fish industry, and are considered opportunistic pathogens of humans causing infections ranging from diarrhea to septicemia. The objective of this study was to determinein silicothe contribution of genomic islands toA. hydrophila. The complete genomes of 17A. hydrophilaisolates, which were separated into two phylogenetic groups, were analyzed using a genomic island (GI) predictor. The number of predicted GIs and their characteristics varied among strains. Strains from group 1, which contains mainly fish pathogens, generally have a higher number of predicted GIs, and with larger size, than strains from group 2 constituted by strains recovered from distinct sources. Only a few predicted GIs were shared among them and contained mostly genes from the core genome. Features related to virulence, metabolism, and resistance were found in the predicted GIs, but strains varied in relation to their gene content. In strains from group 1, O Ag biosynthesis clusters OX1 and OX6 were identified, while strains from group 2 each had unique clusters. Metabolic pathways for myo-inositol, L-fucose, sialic acid, and a cluster encoding QueDEC, tgtA5, and proteins related to DNA metabolism were identified in strains of group 1, which share a high number of predicted GIs. No distinctive features of group 2 strains were identified in their predicted GIs, which are more diverse and possibly better represent GIs in this species. However, some strains have several resistance attributes encoded by their predicted GIs. Several predicted GIs encode hypothetical proteins and phage proteins whose functions have not been identified but may contribute toAeromonasfitness. In summary, features with functions identified on predicted GIs may confer advantages to host colonization and competitiveness in the environment.
- 29 November 2021
- Published by Frontiers Media SA in Frontiers in Microbiology
- Vol. 12
Open AccessPrediction and Analysis in silico of Genomic Islands in Aeromonas hydrophila
Aeromonas are Gram-negative rods widely distributed in the environment. They can cause severe infections in fish related to financial losses in the fish industry, and are considered opportunistic pathogens of humans causing infections ranging from diarrhea to septicemia. The objective of this study was to determinein silicothe contribution of genomic islands toA. hydrophila. The complete genomes of 17A. hydrophilaisolates, which were separated into two phylogenetic groups, were analyzed using a genomic island (GI) predictor. The number of predicted GIs and their characteristics varied among strains. Strains from group 1, which contains mainly fish pathogens, generally have a higher number of predicted GIs, and with larger size, than strains from group 2 constituted by strains recovered from distinct sources. Only a few predicted GIs were shared among them and contained mostly genes from the core genome. Features related to virulence, metabolism, and resistance were found in the predicted GIs, but strains varied in relation to their gene content. In strains from group 1, O Ag biosynthesis clusters OX1 and OX6 were identified, while strains from group 2 each had unique clusters. Metabolic pathways for myo-inositol, L-fucose, sialic acid, and a cluster encoding QueDEC, tgtA5, and proteins related to DNA metabolism were identified in strains of group 1, which share a high number of predicted GIs. No distinctive features of group 2 strains were identified in their predicted GIs, which are more diverse and possibly better represent GIs in this species. However, some strains have several resistance attributes encoded by their predicted GIs. Several predicted GIs encode hypothetical proteins and phage proteins whose functions have not been identified but may contribute toAeromonasfitness. In summary, features with functions identified on predicted GIs may confer advantages to host colonization and competitiveness in the environment.