Fig. 2

Impact of random amplification bias on reads and contigs from saliva viromes. a Fold change of normalised cross-contig abundance (RPKM) between randomly amplified and unamplified viromes are shown. Only those contigs longer than 2 kb with fold change > 50× (green colour) or < 0.02× (red colour) are depicted. Four amplifications were carried out using a GenomiPhi kit with two different DNA template quantities and extension times: 1 ng for 2.5 h (MDA_G1); 1 ng for 10 h (MDA_G2); 10 pg for 3.5 h (MDA_G3); and 10 pg for 10 h (MDA_G4). Amplifications with a TruePrime™ kit were performed from 1 ng for 2.5 h (MDA_T1) and 10 pg for 3.5 h (MDA_T2). SISPA amplifications were carried out with a single primer (FR26RV-12N; SISPA1) or by pooling the amplification products obtained with three different primers (FR26RV-12N, K-12N, and 454-A-12N; SISPA2). b Relative abundance of reads as a function of their average GC content is shown for unamplified and selected randomly amplified viromes. c Fold change of 2577 cross-contigs as a function of their average GC content is shown. Small circular cross-contigs are depicted as blue dots and linear cross-contigs as grey dots. Trend lines obtained by linear regression over two different ranges of %GC are shown