The cluster trees on the top and left margins represent the hierarchical relation between samples and gene counts, respectively.
DESeq was used to normalise read counts for differences in sequencing depth between samples, to robustly estimate the variance of read counts between samples, and to variance-stabilise and log-transform the read count data [ 54].
To calculate CNV distance, the copy state of each sample's call (0,1,2,3+) was assigned to each probe sets positioned within those calls, then pairwise differences between samples were counted for each probe set and divided by the total number of probe sets.
To compare different conditions, we normalize fragment counts between samples, and call for differential contact frequencies using the statistical method DESeq2 adapted from RNA-Seq analysis.
Homozygous differences, which are positions for which two samples are both homozygous for a different base at a single locus, between samples were counted for each pair of samples (Table 4).
Additional file 1: Statistics of RNA-seq, normalized total gene read counts, correlations between samples and expressed genes lists.
Ideally, error rates should be estimated by a combination of approaches, the most widely used being: (1) counting Mendelian inconsistencies in parent offspring pairs; (2) comparing error-prone genotypes with high-quality reference samples; and (3) counting mismatches between duplicated genotypes.
Between-sample read count comparisons showed some evidence of batch effects across N2 embryos, primarily at the level of sequencing lane.
Samples were stained and counted in triplicate.
Additionally, the degree of total methylation for all the sites represented in this assay was compared between cell samples by counting the number of CpG sites falling within a range of β values varying in degree from highly methylated (0.8 1.0) to unmethylated (0 0.2) (Figure 2B).
