Different results with -t rel_ab_w_read_stats and reads_map parameter

Hi there,

I am trying to profile the members in my metagenome data. I tested different types of analysis to perform (-t rel_ab_w_read_stats or reads_map), the other parameters were the same and show up in the attached results files. But I am quite confused with the results. I have attached my results and the intermediate results (tests.bowtie2.txt). Could you please help me have a look?

In the rel_ab_w_read_stats results (test.ra.txt), only species A was identified with 100% abundance. But in the reads_map results (test.read_map.txt), there were quite a lot reads taxonomically assigned, not just species A in the rel_ab_w_read_stats results. Meanwhile, the number of reads assigned to species A was not as high as other assigned species. Why the other species assigned with reads (reads_map result) were not included in the rel_ab_w_read_stats results? Which parameter control the species show up in the rel_ab_w_read_stats results?

Meanwhile, how can I identify more species by just the parameters? At first, I used the default parameters, but the reads were 100% unclassified. Then I tried the --bt2_ps very-sensitive-local, it give me the results above. Are there other methods to identify more species?

Looking forward to your reply.
Thank you.
Yan

test.ra.txt (1.2 KB)
test.read_map.txt (790.2 KB)
tests.bowtie2.txt (253.6 KB)

From the help message:
reads_map: mapping from reads to clades (only reads hitting a marker)
This reports all the clades having at least a read mapping to a marker.

In order to have a species identifies, you need to have at least 1/3 of the markers covered by reads, then a robust average is calculated using a subset of them. You can increase the fraction of those markers by tweaking the stat_q parameter (see Relaxing classification parameters - #2 by fbeghini)