Carolaine da Silva de Araújo
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Carolaine da Silva de Araújo has a total of 14 co-authors. Here are the top co-authors by number of shared publications:
- Magno Junqueira
Laboratório de Espectrometria de Massa Translacional e Neuroproteômica (LEMTEN), Departamento de Bioquímica, Instituto de Química, Universidade Federal do Rio de Janeiro, Rio de Janeiro, RJ 21941-909, Brazil
3 shared publications - Alessandra Leda Valverde
Laboratório de Produtos Naturais (LaProMar), Instituto de Química, Universidade Federal Fluminense, Niterói, RJ, Brazil
2 shared publications - Ana Caroline Silas Rodrigues
Laboratório de Espectrometria de Massa Translacional e Neuroproteômica (LEMTEN), Departamento de Bioquímica, Instituto de Química, Universidade Federal do Rio de Janeiro, Rio de Janeiro, RJ 21941-909, Brazil
2 shared publications - Fábio César Sousa Nogueira
Federal University of Rio de Janeiro
2 shared publications - Christiane Martins de Vasconcellos Silveira
Instituto de Química, Universidade Federal do Rio de Janeiro (UFRJ), Rio de Janeiro, Brasil
1 shared publication - Gabrielle Rodrigues de Souza
Laboratório de Espectrometria de Massas Translacional e Neuroproteômica (LEMTEN), Instituto de Química, Universidade Federal do Rio de Janeiro, Rio de Janeiro, RJ, Brazil
1 shared publication
Latest Publications
- Open Access
Asking the Data: Conversational Interrogation of LC–MS/MS Proteomics Data Sets with an LLM Agent
Proteomics workflows are designed to answer predefined questions, leaving analytical problems that emerge after first results difficult to investigate without additional bioinformatics work. We introduce data set interrogation, a conversational framework in which a researcher uses a large language model (LLM) agent to translate emerging questions into executable tests, check the answers against controls, and keep the code, inputs, and decisions in a dated record. Applied retrospectively to four LC–MS/MS data sets, the framework exposed problems that changed the interpretation of three in-house studies: a search database lacking one of two symbionts, a swapped pair of tandem mass tag (TMT) channels, statistics inflated by non-host proteins, and false deamidation assignments caused by selection of the 13 C isotope peak as monoisotopic; the same question asked of every data set revealed failed or partial cysteine alkylation in all three. A public TMT data set was reanalyzed from raw files in 24 min, a feasibility measure without a human comparison; among spectra identified by both workflows, 99.2% received the same peptide assignment as in Proteome Discoverer, and protein fold changes correlated (r = 0.93). In seeded tests, a purpose-built channel-assignment check detected all 216 between-strain channel swaps and none of 90 within-strain swaps. The framework couples natural-language interaction with executable analyses, explicit controls, and traceable decisions, enabling researchers to investigate questions not anticipated by the original pipeline. These results establish feasibility in a limited retrospective setting with one researcher; broader evaluation across users, models, and data sets is required to determine general performance.
Open AccessAsking the Data: Conversational Interrogation of LC–MS/MS Proteomics Data Sets with an LLM Agent
Proteomics workflows are designed to answer predefined questions, leaving analytical problems that emerge after first results difficult to investigate without additional bioinformatics work. We introduce data set interrogation, a conversational framework in which a researcher uses a large language model (LLM) agent to translate emerging questions into executable tests, check the answers against controls, and keep the code, inputs, and decisions in a dated record. Applied retrospectively to four LC–MS/MS data sets, the framework exposed problems that changed the interpretation of three in-house studies: a search database lacking one of two symbionts, a swapped pair of tandem mass tag (TMT) channels, statistics inflated by non-host proteins, and false deamidation assignments caused by selection of the 13 C isotope peak as monoisotopic; the same question asked of every data set revealed failed or partial cysteine alkylation in all three. A public TMT data set was reanalyzed from raw files in 24 min, a feasibility measure without a human comparison; among spectra identified by both workflows, 99.2% received the same peptide assignment as in Proteome Discoverer, and protein fold changes correlated (r = 0.93). In seeded tests, a purpose-built channel-assignment check detected all 216 between-strain channel swaps and none of 90 within-strain swaps. The framework couples natural-language interaction with executable analyses, explicit controls, and traceable decisions, enabling researchers to investigate questions not anticipated by the original pipeline. These results establish feasibility in a limited retrospective setting with one researcher; broader evaluation across users, models, and data sets is required to determine general performance.
- Open Access
Sample Preparation and Homology-Driven Search Shape the First Holobiont Proteome of the Marine Sponge Viles ophiraphidites (formerly Topsentia ophiraphidites )
Marine sponges are holobionts of host cells, extracellular matrix and dense microbial communities, and most lack reference proteomes. Viles ophiraphidites (formerly Topsentia ophiraphidites ), a Brazilian demosponge, is represented in UniProtKB by a single protein. The study had two phases. In phase 1, four sample-preparation workflows were compared on two specimens in 42 nLC-MS/MS runs: urea/thiourea in-solution digestion, acetone precipitation, S-Trap and GeLC-MS/MS. Every run was analysed with FragPipe and with a homology-driven pipeline that adds three layers to a rescored Sage search: the proteome of the closest sequenced relative ( Halichondria panicea ), MS-BLAST-style alignment of de novo sequence tags, and error-tolerant interpretation of tag flanks. In phase 2, the selected workflow, S-Trap, was applied to four specimens in 12 runs. Both strategies ranked the workflows identically (ρ = 0.98). S-Trap gave the deepest and most specimen-independent proteome, with about 2,100 protein groups per specimen and an 8% difference between specimens, against 21–39% for the other workflows. The homology-driven layers raised assigned non-contaminant spectra by 21.3% in phase 1 and 26.8% in phase 2. They contributed 1,969 protein groups found only by homology and agreed with the database search on 99.5–99.9% of shared spectra. The proteome comprises 4,294 protein groups: 42% sponge and 58% bacterial or archaeal, the latter dominated by Poribacteria. It includes Nitrospira nitrite oxidoreductase detected in every run. The four phase-2 specimens had a similar composition (Bacteria 51–64% and Porifera 35–48% of peptides, Poribacteria alone 27–33%), whereas the two specimens stored long-term in ethanol at 4 °C, which also differed in collection site and storage history, yielded four to five times less evidence than the frozen ones, even though reduction and alkylation had not been effective in the two frozen specimens. S-Trap combined with homology-driven search is a robust strategy for proteomics of non-model holobionts.
Open AccessSample Preparation and Homology-Driven Search Shape the First Holobiont Proteome of the Marine Sponge Viles ophiraphidites (formerly Topsentia ophiraphidites )
Marine sponges are holobionts of host cells, extracellular matrix and dense microbial communities, and most lack reference proteomes. Viles ophiraphidites (formerly Topsentia ophiraphidites ), a Brazilian demosponge, is represented in UniProtKB by a single protein. The study had two phases. In phase 1, four sample-preparation workflows were compared on two specimens in 42 nLC-MS/MS runs: urea/thiourea in-solution digestion, acetone precipitation, S-Trap and GeLC-MS/MS. Every run was analysed with FragPipe and with a homology-driven pipeline that adds three layers to a rescored Sage search: the proteome of the closest sequenced relative ( Halichondria panicea ), MS-BLAST-style alignment of de novo sequence tags, and error-tolerant interpretation of tag flanks. In phase 2, the selected workflow, S-Trap, was applied to four specimens in 12 runs. Both strategies ranked the workflows identically (ρ = 0.98). S-Trap gave the deepest and most specimen-independent proteome, with about 2,100 protein groups per specimen and an 8% difference between specimens, against 21–39% for the other workflows. The homology-driven layers raised assigned non-contaminant spectra by 21.3% in phase 1 and 26.8% in phase 2. They contributed 1,969 protein groups found only by homology and agreed with the database search on 99.5–99.9% of shared spectra. The proteome comprises 4,294 protein groups: 42% sponge and 58% bacterial or archaeal, the latter dominated by Poribacteria. It includes Nitrospira nitrite oxidoreductase detected in every run. The four phase-2 specimens had a similar composition (Bacteria 51–64% and Porifera 35–48% of peptides, Poribacteria alone 27–33%), whereas the two specimens stored long-term in ethanol at 4 °C, which also differed in collection site and storage history, yielded four to five times less evidence than the frozen ones, even though reduction and alkylation had not been effective in the two frozen specimens. S-Trap combined with homology-driven search is a robust strategy for proteomics of non-model holobionts.
- Open Access
Proteome Complexity Scales with Architecture in Human Neural Models
Mass-spectrometry–based proteomics enables high-throughput identification and quantification of proteins, providing molecular insight into neural development and cellular organization. Applying this approach to in vitro systems of increasing architectural complexity, we compared immortalized monolayer cell lines with two tissue-like neural models. Here, we present a comparative proteomic analysis of three human neural models: SH-SY5Y neuroblastoma cells (2D), neurospheres, and cerebral organoids. Protein profiling revealed a stepwise increase in molecular complexity, with enhanced detection of neural-related proteins linked to axon guidance, synapse formation, and GTPase signaling. This trend was most pronounced in tissue-like models, underscoring their suitability for studying neuronal maturation and circuit assembly. Functional enrichment analyses showed progressive acquisition of neurodevelopmental programs, including synaptic vesicle cycling, presynaptic organization, neurotransmitter regulation, and late-stage gliogenesis, accompanied by increased expression of astrocytic and oligodendrocytic markers. Kinase diversity also increased across models, reaching up to 210 regulatory kinases in organoids, many implicated in neural development and degeneration. Gene set enrichment for neurological pathways mirrored this trend, aligning proteomic complexity with disease relevance. Regional brain mapping further indicated that organoids most closely recapitulate the protein architecture of the human CNS. Together, these findings demonstrate that tissue-like neural models provide richer proteomic landscapes that approximate in vivo brain biology and support the application of integrative proteomics in neuroscience and translational research.
Open AccessProteome Complexity Scales with Architecture in Human Neural Models
Mass-spectrometry–based proteomics enables high-throughput identification and quantification of proteins, providing molecular insight into neural development and cellular organization. Applying this approach to in vitro systems of increasing architectural complexity, we compared immortalized monolayer cell lines with two tissue-like neural models. Here, we present a comparative proteomic analysis of three human neural models: SH-SY5Y neuroblastoma cells (2D), neurospheres, and cerebral organoids. Protein profiling revealed a stepwise increase in molecular complexity, with enhanced detection of neural-related proteins linked to axon guidance, synapse formation, and GTPase signaling. This trend was most pronounced in tissue-like models, underscoring their suitability for studying neuronal maturation and circuit assembly. Functional enrichment analyses showed progressive acquisition of neurodevelopmental programs, including synaptic vesicle cycling, presynaptic organization, neurotransmitter regulation, and late-stage gliogenesis, accompanied by increased expression of astrocytic and oligodendrocytic markers. Kinase diversity also increased across models, reaching up to 210 regulatory kinases in organoids, many implicated in neural development and degeneration. Gene set enrichment for neurological pathways mirrored this trend, aligning proteomic complexity with disease relevance. Regional brain mapping further indicated that organoids most closely recapitulate the protein architecture of the human CNS. Together, these findings demonstrate that tissue-like neural models provide richer proteomic landscapes that approximate in vivo brain biology and support the application of integrative proteomics in neuroscience and translational research.