Journal: Nature

Article Title: Developmental convergence and divergence in human stem cell models of autism

doi: 10.1038/s41586-025-10047-5

Figure Lengend Snippet: a , Schematic workflow going from hiPS cells to cortical organoids to sequencing data. The number of hiPS cell lines and individuals (in parentheses) for each form of ASD and controls is indicated. b , Schematic representation of hCO differentiations derived from each hiPS cell line for two forms of ASD (16p11.2 deletion and 22q13.3 deletion). The other forms of ASD can be found in Extended Data Fig. . c , Spearman’s correlation of gene expression between samples from the same time point and form of ASD that were derived either from different lines (red) or from the same line (blue). The sample sizes for each group (number of differentiations) are as follows: day 25: control lines, n = 46; 15q13.3del, n = 7; 16p11.2del, n = 5; 16p11.2dup, n = 4; 22q11.2del, n = 13; 22q13.3del, n = 11; idiopathic, n = 12; PCDH19, n = 2; SHANK3, n = 2; and Timothy syndrome, n = 3. Day 50: control lines, n = 53; 15q13.3del, n = 7; 16p11.2del, n = 6; 16p11.2dup, n = 4; 22q11.2del, n = 15; 22q13.3del, n = 12; idiopathic, n = 15; PCDH19 , n = 2; SHANK3, n = 2; and Timothy syndrome, n = 3. Day 75: control lines, n = 54; 15q13.3del, n = 7; 16p11.2del, n = 6; 16p11.2dup, n = 4; 22q11.2del, n = 23; 22q13.3del, n = 12; idiopathic, n = 15; PCDH19, n = 2; SHANK3, n = 2; and Timothy syndrome, n = 2. Day 100: control lines, n = 50; 15q13.3del, n = 6; 16p11.2del, n = 6; 16p11.2dup, n = 4; 22q11.2del, n = 15; 22q13.3del, n = 12; idiopathic, n = 15; PCDH19, n = 2; SHANK3, n = 2; and Timothy syndrome, n = 2. Boxplots show: centre, median; lower hinge, 25% quantile; upper hinge, 75% quantile; lower whisker, smallest observation greater than or equal to lower hinge −1.5× interquartile range; upper whisker, largest observation less than or equal to upper hinge +1.5× interquartile range. d , Genes within the CNVs are downregulated in deletions and upregulated in duplications, as exemplified by 16p11.2 deletion (del) and duplication (dup) and 15q13.3 deletion. 16p11.2del, 61.5% of genes; 16p11.2dup, 50% of genes and 15q13.3del, 52.9% of genes. *dreamlet P values of less than 0.005. Top illustration adapted from ref. , Springer Nature America; illustrations in the IP–MS and CRISPRi panels created in BioRender. Geschwind, D. (2025) ( https://biorender.com/m2jkj03 ).

Article Snippet: To control for technical variation due to the sequencing and library prep we calculated the principal components of the Picard sequencing metrics ( http://broadinstitute.github.io/picard/ ) using the CollectAlignmentSummaryMetrics, CollectRnaSeqMetrics and MarkDuplicates modules, and included them in our model. To infer genetic ancestry, we called single-nucleotide polymorphisms (SNPs) from the aligned reads using the GATK (v.3.3) Haplotype caller .

Techniques: Sequencing, Derivative Assay, Gene Expression, Control, Whisker Assay, Protein-Protein interactions

Journal: Nature

Article Title: Developmental convergence and divergence in human stem cell models of autism

doi: 10.1038/s41586-025-10047-5

Figure Lengend Snippet: Whole genome sequencing depth in ( a ) 16p11.2 ( b ) 15q13.3 ( c ) 22q11.2 and ( d ) 22q13.3 loci averaged over 10 kb windows. ( e ) Mutational forms not shown in main figure showing genes inside the CNV locus or genes carrying the point mutations (marked in blue). Genes significantly differentially expressed over all time points are denoted by asterisks (dreamlet p-values of p < 0.005). The percent of DE genes in the CNV across the 4 time points is as follows: 22q13del - 77.3% of genes, 22q11del - 88.9% of genes.

Article Snippet: To control for technical variation due to the sequencing and library prep we calculated the principal components of the Picard sequencing metrics ( http://broadinstitute.github.io/picard/ ) using the CollectAlignmentSummaryMetrics, CollectRnaSeqMetrics and MarkDuplicates modules, and included them in our model. To infer genetic ancestry, we called single-nucleotide polymorphisms (SNPs) from the aligned reads using the GATK (v.3.3) Haplotype caller .

Techniques: Sequencing

Journal: Nature

Article Title: Developmental convergence and divergence in human stem cell models of autism

doi: 10.1038/s41586-025-10047-5

Figure Lengend Snippet: ( a ) Overlaps between differentially expressed genes at all time points and risk genes for ASD from either SFARI or from large scale whole exome (WES) sequencing studies , as well as with neurodevelopmental disorders (NDDs) and intellectual disability (ID) risk genes . Colour represents the OR (two-sided Fishers exact test) and the size of the point represented the -log 10 FDR. Only positive significant overlaps (OR > 1 and FDR < 0.05) are shown. ( b ) Select gene ontology (GO) terms enriched in upregulated (red) or downregulated (blue) genes in each of the ASD forms at day 25 (top) or day 100 (bottom). Statistics derived from GSEA. ( c ) Distribution of normalized expression for meta analysis significant example genes at day 100. Boxplots show: centre, median, lower hinge – 25% quantile, upper hinge – 75% quantile, lower whisker – smallest observation greater than or equal to lower hinge –1.5× interquartile range, upper whisker – largest observation less than or equal to upper hinge +1.5× interquartile range. Number of samples (differentiations) for each form: Control – 50, 15q13.3del – 6, 16p11.2del – 6, 16p11.2dup – 4, 22q11.2del – 15, 22q13.3del – 12, Idiopathic – 15, PCDH19 – 2, SHANK3 – 2, Timothy Syndrome – 2. Asterisks denote dreamlet derived p-values: +p < 0.005,* FDR < 0.05, ** FDR < 0.01, *** FDR < 0.005. ( d ) Select GO terms enriched in upregulated or downregulated meta-analysis significant genes Colour corresponds to normalized enrichment score (NES). Point size corresponds to the level of significance (−log 10 (FDR)). Statistics derived from GSEA.

Article Snippet: To control for technical variation due to the sequencing and library prep we calculated the principal components of the Picard sequencing metrics ( http://broadinstitute.github.io/picard/ ) using the CollectAlignmentSummaryMetrics, CollectRnaSeqMetrics and MarkDuplicates modules, and included them in our model. To infer genetic ancestry, we called single-nucleotide polymorphisms (SNPs) from the aligned reads using the GATK (v.3.3) Haplotype caller .

Techniques: Sequencing, Derivative Assay, Expressing, Whisker Assay, Control

Journal: Nature

Article Title: Developmental convergence and divergence in human stem cell models of autism

doi: 10.1038/s41586-025-10047-5

Figure Lengend Snippet: ( a ) Schema of CROP-Seq vector. ( b ) Example of gating for flow cytometry. Left shows a set of hNPCs not expressing virus used for gating compared with hNPCs expressing both dCas9-Krab (GFP) and Crop-Seq-opti-dsRed (TdTomato). Box in the upper right shows double positive cells selected for downstream analysis (~20% of live cell fraction). ( c ) QC measures for CRISPRi single cell libraries organized by gRNA negative (no gRNA detected in sequencing), Control gRNA positive, or Target gRNA positive. ( d ) Number of gRNA UMIs per cell. X-axis shows cut-off of 10 UMIs used to consider a cell gRNA positive (left). Number of of unique gRNAs expressed in each cell (x-axis). Only cells expressing gRNA for a single gene-target are retained for downstream analyses. ( e ) Single cell UMAPs of markers used to differentiate cycling and non-cyling NPCs ( f ) Single cell UMAP coloured by Target-gene presence ( g ) Proportion of cells uniquely expressing gRNAs barcodes for each target within each experiment (n = 18 per Gene comprised of 6 technical replicates, 3 gRNAs barcodes per target, for Controls n = 108: 18 gRNAs across 6 technical replicates). TP53 show increased proportion compared to non-targeting controls (NTCs). Boxplots a-c show: centre, median, lower hinge – 25% quantile, upper hinge – 75% quantile, lower whisker – smallest observation greater than or equal to lower hinge –1.5× interquartile range, upper whisker – largest observation less than or equal to upper hinge +1.5× interquartile range. ***Family wise corrected p < 0.005, Dunnets post-hoc contrast of linear model.

Article Snippet: To control for technical variation due to the sequencing and library prep we calculated the principal components of the Picard sequencing metrics ( http://broadinstitute.github.io/picard/ ) using the CollectAlignmentSummaryMetrics, CollectRnaSeqMetrics and MarkDuplicates modules, and included them in our model. To infer genetic ancestry, we called single-nucleotide polymorphisms (SNPs) from the aligned reads using the GATK (v.3.3) Haplotype caller .

Techniques: Plasmid Preparation, Flow Cytometry, Expressing, Virus, Single Cell, Sequencing, Control, Whisker Assay