Fig. 3: Airqtl enables efficient cell type-specific sceQTL mapping, comprehensive benchmarking, and objective optimization. | Nature Communications

Fig. 3: Airqtl enables efficient cell type-specific sceQTL mapping, comprehensive benchmarking, and objective optimization.

From: Airqtl dissects cell state-specific causal gene regulatory networks with efficient single-cell eQTL mapping

Fig. 3: Airqtl enables efficient cell type-specific sceQTL mapping, comprehensive benchmarking, and objective optimization.The alternative text for this image may have been generated using AI.

a, b Running time (left Y) and computing cost (right Y) for cell type-specific sceQTL mapping on small-scale (a, defaults to \({n}_{{\mathsf{SNP}}}=200\), \({n}_{{\mathsf{gene}}}=1,000\), \({n}_{{\mathsf{donor}}}=50\), \({n}_{{\mathsf{cell}}}=2,000\)) and large-scale (b) defaults to \({n}_{{\mathsf{SNP}}}=200,000\), \({n}_{{\mathsf{gene}}}=2,000\), \({n}_{{\mathsf{donor}}}=100\), \({n}_{{\mathsf{cell}}}=50,000\)) datasets with varying dimensions. Color: method. c, d Low bias (c) and variance (d) in cell type-specific sceQTL effect size estimation by airqtl, stratified by expression quartiles (X). Dashed line: perfect performance. e Airqtl’s scalability enabled benchmarking and optimization of P value calibration for cell type specificity. Null P value histograms are shown for non-eQTLs (left) and non-specific eQTLs (right) before (raw) and after (calibrated) calibration. Shades in histograms represent error bars (\(3\sqrt{N}\), where N is the number of entries in each bin). Dashed line: perfect performance (standard uniform distribution).

Back to article page