Journal: Science Advances

Article Title: Systems biology framework for the rational design of operational conditions for in vitro/in vivo translation of tissue models

doi: 10.1126/sciadv.aef7756

Figure Lengend Snippet: ( A ) Bulk RNA-seq data are used to translate between clinical samples (in vivo) with phenotypic annotations and an MPS model of disease (in vitro). ( B ) The MASLD score (MAS) and fibrosis stage are clinically relevant metrics of disease that correlate and need to be modeled jointly. ( C ) A PLSR model predicts both phenotypes from bulk transcriptomics and performs significantly better in 10-fold cross-validation than random or shuffled models. For all comparisons, a two-sided unpaired Wilcoxon test was used. In all box plots, the centerline denotes the median, the bounds of the box denote the first and third quantiles, and the whiskers denote points not being further from the median than 1.5 × interquartile range. ( D ) The PLSR model with eight LVs fits the full clinical dataset well and achieves good correlations with the measured phenotypes. ( E ) Qualitative trends in the phenotypic scores are visible when projecting the clinical data onto the first two LVs of the PLSR model.

Article Snippet: Purified RNA was sent to Plasmidsaurus for bulk RNA-seq using the Illumina sequencing technology.

Techniques: RNA Sequencing, In Vivo, In Vitro, Transcriptomics, Biomarker Discovery

Journal: Science Advances

Article Title: Systems biology framework for the rational design of operational conditions for in vitro/in vivo translation of tissue models

doi: 10.1126/sciadv.aef7756

Figure Lengend Snippet: ( A ) Schematic of validation liver triculture spheroid experiments. Hepatocytes, Kupffer cells, and hepatic stellate cells were seeded in a 10:1:1 ratio in alginate wells and cultured for 11 days. Treatments were initiated on day 3 (D3) and replenished with each medium change. ( B and C ) Quantification of spheroid phenotypes shows that TGFβ1 (10 ng/ml) decreased steatosis, as measured by normalized BODIPY signal (B), while increasing fibrosis, measured by α-SMA in vimentin-positive areas (C). Cotreatment with IFN-α (1000 U/ml) reduced lipid content further and restored α-SMA to baseline expression. a.u., arbitrary units. ( D ) Bulk RNA-seq performed on parallel spheroid cultures reveals strong transcriptional differences associated with the experimental treatments, as evidenced by PCA. ( E ) Pathway activity inference confirms a TGFβ signature associated with PC1 and a JAK-STAT signature associated with PC2. ( F ) Projection of the gene expression data onto TC1 and extra LV 1 preserves clustering by experimental condition. Imaging data are presented as a ratio normalized to the median of the control group. Each point represents an individual spheroid. In all panels, asterisks indicate statistical significance level defined as follows: **** P ≤ 10 –4 , *** P ≤ 10 –3 , and ns for P > 0.05.

Article Snippet: Purified RNA was sent to Plasmidsaurus for bulk RNA-seq using the Illumina sequencing technology.

Techniques: Biomarker Discovery, Cell Culture, Expressing, RNA Sequencing, Activity Assay, Gene Expression, Imaging, Control

Journal: Molecular Therapy. Nucleic Acids

Article Title: Enhancing protein expression in humans through codon optimization with transformer and contrastive learning

doi: 10.1016/j.omtn.2026.102991

Figure Lengend Snippet: Experimental comparison of codon-optimized constructs in HEK293T cells (A) Western blot analysis of HEK293T cells transfected with wild-type or codon-optimized EMG1 , JNK1 , and CREB1 constructs generated by ExpOptimizer, GenSmart, or COformer. Protein expression was detected using an anti-His antibody, with GAPDH as a loading control. (B) Quantification of protein expression normalized to GAPDH and shown as fold change relative to wild-type. Data represent mean ± SD from three independent experiments. Statistical significance was assessed using one-way ANOVA followed by Tukey’s multiple comparison test. ∗ p < 0.05, ∗∗ p < 0.01 vs. wild-type; # p < 0.05, ## p < 0.01 vs. ExpOptimizer; & p < 0.05, && p < 0.01 vs. GenSmart. (C) Relative transcript abundance measured by RNA-seq 24 h post-transfection and normalized to GAPDH . Expression values are shown as fold change relative to wild type.

Article Snippet: Three micrograms of purified total RNA for each sample was shipped for genome-wide RNA sequencing (Plasmidsaurus).

Techniques: Comparison, Construct, Western Blot, Transfection, Generated, Expressing, Control, RNA Sequencing

Journal: Molecular Therapy. Nucleic Acids

Article Title: Enhancing protein expression in humans through codon optimization with transformer and contrastive learning

doi: 10.1016/j.omtn.2026.102991

Figure Lengend Snippet: Benchmarking COformer against commercial tools and learning-based models on a held-out test set COformer was compared with ExpOptimizer, GenSmart, GeneArt, ICOR, and CodonTransformer using identical held-out protein inputs. Sequence-level descriptors included (A) CAI, (B) overall GC fraction, (C) GC3 fraction, (D) uridine fraction, and (E) tAI. For (A–E), each distribution represents sequence-level values calculated for individual held-out protein inputs. Violin width reflects the density of observations, and internal lines indicate the 25th percentile, median, and 75th percentile. (F) Predicted RNA secondary-structure MFE was computed using ViennaRNA. For the boxplot, the center line indicates the median, the box spans the interquartile range, and whiskers extend to the most extreme values within 1.5 times the interquartile range.

Article Snippet: Three micrograms of purified total RNA for each sample was shipped for genome-wide RNA sequencing (Plasmidsaurus).

Techniques: Sequencing