Journal: bioRxiv

Article Title: BioGAIP: A Scalable, User-Friendly and Robust LLM-Powered Multi-Agent System for Automated Bioinformatics Tasks

doi: 10.64898/2026.05.16.720484

Figure Lengend Snippet: (a) Flowchart of BioGAIP analyzing RNA-seq data. The expected endpoint of this task is the generation of a list of differentially expressed genes (DEGs). During the successive process, human intervention can be introduced to prompt BioGAIP for specific downstream analyses. (b) Flowchart of BioGAIP analyzing ATTSS events. The endpoint of this task is to generate a list of differential ATTSS events. The LLM model employed in the analysis processes for (a) is qwen3-max and for (b) is grok-4-fast-reasoning. Titles in green boxes indicate normally executed steps. Titles in brown boxes indicates steps where BioGAIP encountered an error. Titles in blue boxes indicate troubleshooting steps performed by BioGAIP, with blue text showing the specific error cause(s) (when present). Titles in dark green boxes indicate steps that involved manual prompt input by the user. Arrows denote the direction of the workflow, troubleshooting flows are shown with dashed lines.

Article Snippet: The single-cell RNA-seq reference package (refdata-gex-GRCh38-2024-A.tar.gz) from the Cell Ranger website ( https://www.10xgenomics.com/support/software/cell-ranger/downloads ).

Techniques: RNA Sequencing

Journal: bioRxiv

Article Title: BioGAIP: A Scalable, User-Friendly and Robust LLM-Powered Multi-Agent System for Automated Bioinformatics Tasks

doi: 10.64898/2026.05.16.720484

Figure Lengend Snippet: (a) Volcano plot of DEGs in bulk RNA-seq data of met-associated primary SCLC and never-met primary SCLC. (b) UMAP of cell annotation and FOXA2 expression level in each component based scRNA-seq. (c) Correlation plot of the top 100 highly expressed genes in FOXA2 + vs. FOXA2 - cells from Kawasaki et al. gene set, relative to FOXA2 expression defined by Kawasaki et al . (d,e) Density plot of ASCL1 ChIP-seq gene loci at the FOXA2 and PROX1 gene loci in two SCLC cell lines (H1836 and SHP-77). (f) Density plot at the FOXA2 locus of ATAC-seq data derived from ASCL1 + FOXA2 + PDX vs ASCL1 + FOXA2 - PDX tumors.

Article Snippet: The single-cell RNA-seq reference package (refdata-gex-GRCh38-2024-A.tar.gz) from the Cell Ranger website ( https://www.10xgenomics.com/support/software/cell-ranger/downloads ).

Techniques: RNA Sequencing, Expressing, ChIP-sequencing, Derivative Assay

Journal: bioRxiv

Article Title: BioGAIP: A Scalable, User-Friendly and Robust LLM-Powered Multi-Agent System for Automated Bioinformatics Tasks

doi: 10.64898/2026.05.16.720484

Figure Lengend Snippet: (a) Volcano plot of differential ATTSS events. (b) RNA-seq density plots showing the high distal TSS usage of RAB35 in met-associated primary tumors and never-met primary tumors. (c) The Venn diagram between different ATTSS events and differential expression genes. (d) The effect of different ATTSS events on gene coding region. (e) The comparison of distal TSS usage of RAB35 between never-met primary and met-associated primary tumor. (f) The comparison of gene expression of RAB35 between never-met primary and met-associated primary tumor. Statistical significance of distal TSS usage was assessed using the DATTS , statistical significance of gene expression was assessed using the DESeq2 . The analysis of (c) was completed by human experts based on the output of BioGAIP.

Article Snippet: The single-cell RNA-seq reference package (refdata-gex-GRCh38-2024-A.tar.gz) from the Cell Ranger website ( https://www.10xgenomics.com/support/software/cell-ranger/downloads ).

Techniques: RNA Sequencing, Quantitative Proteomics, Comparison, Gene Expression

Journal: bioRxiv

Article Title: BioGAIP: A Scalable, User-Friendly and Robust LLM-Powered Multi-Agent System for Automated Bioinformatics Tasks

doi: 10.64898/2026.05.16.720484

Figure Lengend Snippet: (a) Heatmap of differential ATTSS events based on DTUI values. (b) RNA-seq density plots show the high distal TSS usage of AP4E1 in two met-associated primary tumors and two never-met primary tumors. (c) The comparison of distal TSS usage of AP4E1 between never-met primary and met-associated primary tumor. (d) The comparison of gene expression of AP4E1 between never-met primary and met-associated primary tumor. (e-f) The comparison of gene expressions of RAB35 (e) and AP4E1 (f) between normal and sample tumor in GES60052. Statistical significance of distal TSS usage was assessed using the DATTS , statistical significance of gene expression was assessed using the DESeq2 .

Article Snippet: The single-cell RNA-seq reference package (refdata-gex-GRCh38-2024-A.tar.gz) from the Cell Ranger website ( https://www.10xgenomics.com/support/software/cell-ranger/downloads ).

Techniques: RNA Sequencing, Comparison, Gene Expression

Journal: Nature Aging

Article Title: Proteomic signatures of the APOE ε 4 and APOE ε 2 genetic variants and Alzheimer’s disease

doi: 10.1038/s43587-026-01123-0

Figure Lengend Snippet: We performed different targeted analyses based on the available data for each cohort. The GNPC cohort ( N = 3,289 individuals; plasma, SomaLogic 7K) was used to identify proteins associated with APOE4 or APOE2 and their roles in clinical AD (CU versus AD dementia). Replication was performed in BioFINDER-2 (plasma, SomaLogic 7K). Further validation was conducted in CSF using ADNI (SomaLogic 7K, TMT-MS) and BioFINDER-2 (OLINK). To enable a systematic four-way comparison of APOE -associated proteomic signatures across tissue (plasma versus CSF) and platform (SomaLogic versus OLINK), we incorporated plasma OLINK data from the population-based cohort UKBB. Longitudinal APOE –protein associations were evaluated in PPMI (CSF, OLINK), providing exploratory insights into temporal stability independent of Aβ pathology. Specifically, differential abundance analysis identified APOE ( APOE4 or APOE2 )-associated proteins and AD diagnosis (or Aβ status)-associated proteins, resulting in different groups of proteins specifically associated with APOE or AD or jointly associated with both. Proteins associated with APOE were further tested and categorized according to whether they showed stronger evidence for upstream versus downstream mediation using two mediation models, upstream mediation model: APOE → protein → AD diagnosis or Aβ status, and downstream mediation model: APOE → AD diagnosis or Aβ status → protein. Stratified analyses by AD diagnosis or Aβ status were conducted to investigate in depth the changes of APOE –protein associations. Age-stratified analyses were performed only in CU or Aβ − individuals to investigate how early APOE –protein associations change with age. APOE4 and APOE2 effects were analyzed separately, with direct comparisons across six genotype groups (ε2/ε2, ε2/ε3, ε2/ε4, ε3/ε3, ε3/ε4, ε4/ε4) to evaluate allele dominance. Functional annotation included cell-type enrichment, GO enrichment, BINNs-enriched Reactome pathway analysis, and protein–protein interaction (PPI) analysis. Associations with 5 AD phenotypes were evaluated to link APOE -related proteins to disease features, including tau-PET, Aβ-PET, cortical thickness, cognition (Mini-Mental State Examination (MMSE) and modified Preclinical Alzheimer Cognitive Composite (mPACC)). To assess the robustness of key findings and support their central relevance, we evaluated genetic evidence from AD-associated SNPs in coding genes and examined spatial transcriptomic co-expression with APOE in the human brain. Matching superscript numbers indicate which analyses were conducted in which cohorts. A summary of the cohort analysis can be found in Supplementary Fig. . Figure created in BioRender; Lu, L. https://BioRender.com/vvnh8bs (2026).

Article Snippet: For the Human Protein Atlas 81-cell-type reference, enrichment was assessed using expression-weighted cell-type enrichment (EWCE ) R package (v1.18.0).

Techniques: Clinical Proteomics, Biomarker Discovery, Comparison, Functional Assay, Modification, Expressing

Journal: Nature Aging

Article Title: Proteomic signatures of the APOE ε 4 and APOE ε 2 genetic variants and Alzheimer’s disease

doi: 10.1038/s43587-026-01123-0

Figure Lengend Snippet: a , Volcano plot for proteins associated with APOE2 without adjusting for AD diagnosis ( N = 2,012 individuals, linear models adjusted for age, sex, mean protein level, and cohorts), with red representing significant association after FDR correction. On the y axis, −log 10 (FDR) above 300 was set to 300 for a better visualization. b , Volcano plot shows proteins associated with AD diagnosis without adjusting for APOE2 ( N = 2,012 individuals, linear models adjusted for age, sex, mean protein level, and cohorts), with blue representing significant association after FDR correction. c , UpSet plot shows the number of proteins associated with APOE2 or clinical AD with or without adjusting for each other, blue indicating AD-specific proteins, red indicating APOE2- specific proteins, black indicating the number of proteins independently associated with both. d , Scatter plot shows APOE2 ’s effect size on proteins without adjusting AD diagnosis in the whole cohort ( x axis, N = 2,012 individuals) versus in CU subgroup ( y axis, N = 1,446 individuals) for each protein. Effect sizes were derived from linear models adjusted for age, sex, mean protein level, and cohorts. Spearman correlation was assessed with a two-sided test. Proteins highlighted in red are significantly associated with APOE2 in both the whole cohort and the CU subgroup. e , VPS29 protein levels are plotted against age and stratified by APOE2 status (ε2 + versus ε3/ε3) and clinical diagnosis (CU versus AD). Each line represents a group. Solid lines indicate LOESS (locally estimated scatterplot smoothing)-fitted mean VPS29 levels across age, and shaded bands indicate the 95% confidence intervals around the fitted mean. f , BINN-enriched pathway analysis for early dysregulated proteins in APOE2 carriers; the darker the color, the more important the protein or pathway in predicting AD dementia diagnosis. More features are hidden in the sink for a better visualization. g , The heat map summarizes mediation effects and statistical significance for APOE2 -associated proteins (75 in total were involved in either pathway) across the two mediation pathways: the upstream pathway (top row; APOE2 → protein → AD, red label) and the downstream pathway (bottom row; APOE2 → AD → protein, blue label). Cell colors represent the proportion of mediation. Protein labels on the x axis are color coded by the dominant mediation direction, with red indicating stronger upstream mediation and blue indicating stronger downstream mediation. Bold labels denote complete mediation within the dominant pathway, whereas nonbold labels denote partial mediation. Asterisks indicate statistical significance (*+ FDR-corrected significance). For clarity, only selected proteins with the largest mediation effects from each pathway are shown; full results are provided in the . h , Subdivision of APOE2 -associated proteins based on the association between proteins, APOE2 , and AD. Note that overlapping proteins are preferentially assigned to mediation categories. i , The LDA score plot shows the projection score of each group of subdivided proteins in the discriminant direction. Proteins are colored by their assigned groups. j , The integrative matrix summarizes differential regulation (red for upregulated and blue for downregulated proteins in APOE2 carriers), cell-type enrichment based on scaled RNA expression, and functional characterization of each protein. Cell types from the ROSMAP atlas are labeled in black on the x axis, while those from the BBB atlas are labeled in red. Gray boxes indicate nominal significance ( P < 0.05), and black boxes indicate FDR-corrected significance ( P FDR < 0.05) in cell-type enrichment analysis. GO biological process terms associated with each protein are grouped into broader representative categories; small red boxes indicate the involvement of a given protein in the corresponding process. Cell-type enrichment and GO enrichment analyses were one-sided, with Benjamini–Hochberg adjustment for multiple comparisons. PPIs are annotated using STRING database interactions with a confidence score ≥ 0.7. The number of interactions per protein is shown as a heat map, and direct interactions between proteins are represented by lines, color coded according to their assigned cluster. See Source Data Fig. for detailed statistical summary.

Article Snippet: For the Human Protein Atlas 81-cell-type reference, enrichment was assessed using expression-weighted cell-type enrichment (EWCE ) R package (v1.18.0).

Techniques: Biomarker Discovery, Derivative Assay, RNA Expression, Functional Assay, Labeling

Journal: Nature Aging

Article Title: Proteomic signatures of the APOE ε 4 and APOE ε 2 genetic variants and Alzheimer’s disease

doi: 10.1038/s43587-026-01123-0

Figure Lengend Snippet: a , Volcano plot for proteins associated with APOE4 without adjustment for AD diagnosis ( N = 2,864 individuals, linear models adjusted for age, sex, mean protein level, and cohorts), with red representing significant association after FDR correction. On the y axis, −log 10 (FDR) above 300 was set to 300 for a better visualization. b , Volcano plot shows proteins associated with AD diagnosis without adjustment for APOE4 ( N = 2,864 individuals, linear models adjusted for age, sex, mean protein level, and cohorts), with blue representing significant association after FDR correction. c , UpSet plot shows the number of proteins associated with APOE4 or AD with or without adjusting for each other, blue indicating AD-specific associated proteins, red indicating APOE4- specific associated proteins, black indicating the number of proteins jointly associated with both. d , Scatter plot shows APOE4 ’s effect size on protein without adjusting AD diagnosis in the whole cohort ( x axis, N = 2,864 individuals) versus in CU subgroup ( y axis, N = 1,751 individuals) for each protein. Effect sizes were derived from linear models adjusted for age, sex, mean protein level, and cohorts. Spearman correlation was assessed with a two-sided test. Red represents proteins associated with APOE4 in both the whole cohort and in the CU group. e , BINN-enriched Reactome pathway analysis for proteins associated with APOE4 in both the whole cohort and in CU. The darker the dot, the more important the protein and the pathway in the deep learning model predicting AD dementia diagnosis. More features are hidden in the sink for a better visualization. f , The scatter plot shows the effect of APOE4 on proteins in CU subgroup ( x axis, N = 1,751 individuals) versus the effect of AD on proteins in ε3/ε3 carriers ( y axis, N = 1,679 individuals). Only proteins associated with APOE4 in CU individuals and with AD diagnosis in ε3/ε3 carriers are visualized. Red indicates the same effect direction, while blue indicates an opposite effect direction. g , The heat map summarizes mediation effects and statistical significance for APOE4 -associated proteins (216 in total were involved in either pathway) across the two mediation pathways: the upstream pathway (top row; APOE4 → protein → AD, red label) and the downstream pathway (bottom row; APOE4 → AD → protein, blue label). Cell colors represent the proportion of mediation. Protein labels on the x axis are color coded by the dominant mediation direction, with red indicating stronger upstream mediation and blue indicating stronger downstream mediation. Bold labels denote complete mediation within the dominant pathway, whereas nonbold labels denote partial mediation. Asterisks indicate statistical significance (*+ FDR-corrected significance). For clarity, only selected proteins with the largest mediation effects from each pathway are shown; full results are provided in the . h , Venn plot shows the number of proteins in each category. Note that overlapping proteins are preferentially assigned to mediation categories. i , The linear discriminant score plot shows the projection score of all tested proteins in the discriminant direction. Proteins are colored by their assigned groups. j , The integrative matrix summarizes differential regulation (red for upregulated and blue for downregulated proteins in APOE4 carriers), cell-type enrichment based on scaled RNA expression, and functional characterization of each protein. Cell types from the ROSMAP atlas are labeled in black on the x axis, while those from the BBB atlas are labeled in red. Gray boxes indicate nominal significance ( P < 0.05), and black boxes indicate FDR-corrected significance ( P FDR < 0.05) in cell-type enrichment analysis. GO biological process terms associated with each protein were grouped into broader representative categories; small red boxes indicate the involvement of a given protein in the corresponding process. Cell-type enrichment and GO enrichment analyses were one-sided, with Benjamini–Hochberg adjustment for multiple comparisons. PPIs are annotated using STRING database interactions with a confidence score ≥ 0.7. The number of interactions per protein is shown as a heat map, and direct interactions between proteins are represented by lines, color coded according to their assigned cluster. See Source Data Fig. for detailed statistical summary.

Article Snippet: For the Human Protein Atlas 81-cell-type reference, enrichment was assessed using expression-weighted cell-type enrichment (EWCE ) R package (v1.18.0).

Techniques: Biomarker Discovery, Derivative Assay, RNA Expression, Functional Assay, Labeling