Fig 1.
Schematic overview of Ampli1™ LowPass approach.
DNA is amplified through primers complementary to Ampli1™ WGA universal adapters through a single PCR reaction. Primers incorporate Ion Torrent™-compatible adapter sequences and barcodes. Libraries are then pooled and subjected to standard processing for sequencing on PGM or Ion S5 platforms.
Fig 2.
Effect of normalization on read counts distribution.
a) Scatter plot of read counts, normalized on 1 million of reads, versus GC content in 500 Kbp bins obtained by sequencing of a single WBC; number of MseI fragments per bin is plotted b) respect to GC content and c) along the 22 autosomes; scatter plots of read counts in a single WBC versus number of MseI fragments per bin, weighted on per-fragment probabilities, before d) and after e) GC normalization, three standard deviations are used to discriminate outliers (red dots); f) GC-normalized read counts plotted along the 22 autosomes.
Fig 3.
Performance of CNA calling in amplified vs. non-amplified DNA in 4 aberrant cell lines.
Data obtained by low-pass WGS (0.5-1M reads) of DNA from single cells amplified with Ampli1™ WGA kit were processed for CNA calling. CNAs detected in non-amplified bulk gDNA (20-30M reads) were used as reference. For all the 4 cell lines considered ROC analysis showed an excellent agreement (0.91≤AUC≤0.97) between CNA calls from single cells and bulk gDNA.
Fig 4.
CNA detection by low-pass experiments at different read depths and resolution.
Two cells (a-d & e-h) from cell line NCI1650 were analyzed at different window size/resolutions (a,e = 100Kb; b,f = 200Kb; c,g = 500Kb; d,h = 2,000Kb). A dataset at 3,500,000 reads served as reference for ROC analysis.
Fig 5.
Comparison of LowPass copy number profiles and CNA calling with aCGH.
Example profiles from one single cell of aberrant cell line NCI-H23 generated by Ampli1™ LowPass (a) and aCGH of Ampli1™ amplified DNA (b). In c-p): ROC curves comparing Ampli1™ LowPass CNA calls with aCGH calls from single cell of 6 cell lines of the NCI-H series.
Fig 6.
Determination of single cell ploidy.
Analysis of one single cell from the near-diploid cell line NCI-H23 analyzed using a main ploidy of 2 (red) and 3 (blue): a) copy number profiles along 22 chromosomes; b) copy number levels distribution; c) density estimated by KDE; peaks detected are indicated as dashed vertical lines; d) linear regression of peak values over putative underlying copy numbers: clearly peaks obtained with a main ploidy of 2 better approximate the regression line compared to those obtained at a main ploidy of 3.
Fig 7.
Absolute copy number CNA calling in a single cell of hyperhexaploid cell line NCI-H661.
Plots of copy number profiles along the 22 autosomes expressed as absolute copy numbers. In a) and b) profiles obtained from the same sequencing data with main ploidy parameter set to 2 and 6 respectively. Significant copy number gains and losses are highlighted in red and blue respectively. Clearly a main cell ploidy = 6 provides a better fit of profiles with segmented data (black lines) and improves CNA calling. CNA calls only detected with main ploidy = 6 are shaded in green.
Fig 8.
Cluster analysis of copy number profiles for CTCs and WBCs from 3 patients.
a) single cells (CTCs and WBCs) from a patient affected by prostate cancer; cluster A represents 6 CTCs with small or no differences in copy number profiles; cluster B is formed by WBCs clustering, as expected, on a distinct branch of the tree. b,c) single cells (CTCs and WBCs) from 2 patients affected by lung adenocarcinoma. Values are expressed as fold changes respect to the main ploidy.