Tweet
Hey all,
I am currently working with SE reads (75bp) from 4 independent samples. Sequencing generated approximately 25 million reads per sample (100 million total). My goal is to perform RNA seq to look at gene expression. I am using CLC workbench.
Before I performed the de novo assembly I filtered reads less than 15 bp long and with greater than 2 uncalled bases (N). After performing the de novo assembly on the 100 million reads I get approximately 193000 contigs, with a min size of 100bp.
I have begun performing the RNA seq and the first thing I noticed was that only approximately 30% of the reads from the individual samples are mapping back to the contigs. Am I right in saying that this number is very low? The unmapped reads are a mix of long and short reads, they appear to be high quality, and when I randomly blast them with blastn I get strong matches to sequences from related species. I tried relaxing the mapping parameters some but it only increased the number of reads mapping by 5%.
I am very much beginner when it comes to this technology and what I have learned so far comes from trial and error on my part, as well as a question or two that I have posted on this forum. It seems to me that if you are mapping reads to contigs generated from those reads then the mapping efficiency should be much higher than 30-35%. I will continue to play with the mapping and the de novo assembly to try and improve the mapping but I was hoping someone on here could comment and maybe provide some insight into this.
Thanks!
I am currently working with SE reads (75bp) from 4 independent samples. Sequencing generated approximately 25 million reads per sample (100 million total). My goal is to perform RNA seq to look at gene expression. I am using CLC workbench.
Before I performed the de novo assembly I filtered reads less than 15 bp long and with greater than 2 uncalled bases (N). After performing the de novo assembly on the 100 million reads I get approximately 193000 contigs, with a min size of 100bp.
I have begun performing the RNA seq and the first thing I noticed was that only approximately 30% of the reads from the individual samples are mapping back to the contigs. Am I right in saying that this number is very low? The unmapped reads are a mix of long and short reads, they appear to be high quality, and when I randomly blast them with blastn I get strong matches to sequences from related species. I tried relaxing the mapping parameters some but it only increased the number of reads mapping by 5%.
I am very much beginner when it comes to this technology and what I have learned so far comes from trial and error on my part, as well as a question or two that I have posted on this forum. It seems to me that if you are mapping reads to contigs generated from those reads then the mapping efficiency should be much higher than 30-35%. I will continue to play with the mapping and the de novo assembly to try and improve the mapping but I was hoping someone on here could comment and maybe provide some insight into this.
Thanks!