Assembling Mitochondrial Genome from Single-Cell Data

Approximately 10 000 single-cell G.pseudospinescens organisms were isolated by hand from Gullmar Fjord sediment (Risgaard-Petersen et al., 2006 (link)). After washing, total DNA was extracted using the QIAamp DNA Micro Kit and sequenced by Illumina Genome Analyser II. Using Edena (Hernandez et al., 2008 (link)), 9 950 730 32 nt reads were assembled with parameters m=16 and M=16, which yielded the highest N₅₀ value (N₅₀=170). The raw data and assembly are available from the Gene Expression Omnibus (GEO) under accession number GSE26664. The total DNA of a eukaryote may contain up to three different translation codes (De Grey, 2005 (link)): nuclear, mitochondrial and plastid (if the organism is photosynthetic, but this is not the case for G.pseudospinescens). To avoid mixing these signals, the user can choose to feed individual contigs to FACIL, but this might lead to a bad genetic code prediction due to shortage of data. Thus, we selected those contigs that were likely derived from the G.pseudospinescens mitochondrial genome as follows. The 8456 assembled contigs were queried by BlastX version 2.2.22+ (Camacho et al., 2009 (link)) against all proteins encoded by completely sequenced mitochondria, downloaded from NCBI organelle genome resources (http://www.ncbi.nlm.nih.gov/genomes/GenomesHome.cgi?taxid=2759) on July 28, 2010. Importantly, we used the standard genetic code for this BlastX search in order not to impose a bias in the genetic code on the contigs and our results. The 150 contigs with a high-scoring BlastX hit (E-value ≤0.01) were considered to be of mitochondrial origin (average length 223 nt, median length 191 nt). These sequences are available as ‘example’ input data on the FACIL web server. They contain fragments of mitochondrial genes like cytochrome B and several ATP synthase, cytochrome-c oxidase and NADH dehydrogenase subunits. We found no evidence for multiple copies (e.g. a nuclear and a mitochondrially encoded copy) of the encoded genes after a BlastN search (E-value ≤0.01) of the contigs against themselves.