View Single Post
Old 12-17-2016, 08:41 AM   #402
Location: New York

Join Date: Dec 2016
Posts: 22

For maaping diploid maize data, I concatenated all the chromosomes for a given species (e.g. 10 chr in maize) and used it as the mapping reference. I then mapped the BBDuk-trimmed reads to this reference (genome, not transcriptome) using the default setting of BBMap.
For the maize data generated from B73 (B73 is the sequeced ref genome), the ambig rate rages from 3% to 15%. I saw the variation occurred between biologica rep, not within a pair. For example, the ambig rates of Read 1 and Read 2 of a replicate are close to each other, but the ambig rates of different biologica replica could be various, (12%, 3%, 3%; 11%, 11%, 3%). The variatoin between replica almost made me feel like it was due to the technical issue of the libraries and/or biological nature of maize. I could try to map it to the transcriptome and see if there's any improvement. Other than that I wasn't sure how to trace down, and/or if this should be a concern.
For a different set of maize data with mixed background, as for now I still used the same B73 base reference genome for mapping. The ambig ranges went up to 6% - 46%. The increase was expected given the SNPs/indels present between different cultivars. I also observed variation between biological replicates similar to described above.
On a related note, I also have diploid Arabidopsis data which I also mapped to its own genome (TAIR9 genome fasta). I added a -maxindel=2000 to accomodate the compact size of the genome. The ambig rate on average was lower (mostly below 4%). I did not see that radical variation between replicates either. This is consistent with my guess about the ambig rates in maize was partly due to the nature of maize.
Please let me know if any furtehr details would be helpful for you to diagnose. Thank you as always for your input.
chiayi is offline   Reply With Quote