impact of quality primer on the quality of data

[Cravo_Adri (Adriana Paula Bernardo Cravo)]

Hi, everyone,
I searched in the forum and did not find the answer to my question, maybe you can help me.
I'm sequencing my samples for both 16S and 18S, but they come from a vertebrate, so there was a lot of host sequences in our dataset and we are working now on a blocking primer to decrease this abundance.
We are developed a blocking primer specific to the sequence more similar to our host, according to NCBI, and compared to the results when no blocking primer was used.
This is the quality plot without blocking primer:

and this is the quality plot with blocking primer:

. In the "reverse reads" plot, the first sequences show a weird pattern and my first question is if it's enough to simply remove those sequences, with the p-trim-left-r command in DADA2 (I didn't use cut adapt here) or if this could have an effect on the general quality of the reads.
Additionally, I also compared the total amount of reads per sequencing run (without x with block primer). Without blocking primer:
and with blocking primer:
(I had more samples in the sequencing run with blocking primer, but the samples that were sequenced when no blocking primer was applied were again tested with the blocking primer). The difference in the amount of sequences in "input" and "non-chimeric" is clearly big. I believe this is mostly due to the host sequences that were blocked, however I'll count them and compare in the datasets without and with blocking primer to see if the numbers are similar.
Here comes my second observation. The abundance of other taxons (such as SAR, Fungi etc) also decreased when the blocking primer was used, which, theoretically, should not happen. Here are two plots only for visualization: without blocking primer, SAR plot

and with blocking primer, SAR plot
So now I'm not sure if the use of blocking primer also influences negatively the general quality of my dataset and to which extension it is a good option or not. I'm having problems to define this threshold and it would be good if someone knew if there are other options to check the efficiency of the primer.

Thanks a lot in advance!!!

[colinbrislawn (Colin J Brislawn)]

Hello Adriana,

Let me see if I can answer at least some of your questions.

@Cravo_Adri

I’m sequencing my samples for both 16S and 18S, but they come from a vertebrate, so there was a lot of host sequences in our dataset and we are working now on a blocking primer to decrease this abundance.

This sounds like a cool project. The use of blocking primers is especially interesting!

@Cravo_Adri

it’s enough to simply remove those sequences, with the p-trim-left-r command in DADA2

Yes, that's a good idea.

@Cravo_Adri

or if this could have an effect on the general quality of the reads.

The blocking primer should not effect resulting read quality... I don't think. I'm a little surprised that your quality looks so different.

@Cravo_Adri

The difference in the amount of sequences in “input” and “non-chimeric” is clearly big. I believe this is mostly due to the host sequences that were blocked, however I’ll count them and compare in the datasets without and with blocking primer to see if the numbers are similar.

If the blocking primers were to... uh... block reads from entering the data set, then you would see this in the input column only. In your blocking data set, I notice a big decrease in your filtered column, meaning that reads were still in the input, but were removed due to low quality. The quality in your blocking data set is low, so I'm not sure if the run quality or blocking primers is causing the difference.

Maybe your blocking sequencing run was bad.
Did you consider randomly assigning blocked and unblocked samples too each of the sequencing runs?

---

@Cravo_Adri

The abundance of other taxons (such as SAR, Fungi etc) also decreased when the blocking primer was used, which, theoretically, should not happen.

Those two plots look pretty similar to me; I see a lot of Alveolata, and a few Rhizaria and Stramenopiles in a few of the samples.

Because blocked and unblocked samples were on different runs, it's hard to know if changes are due to the blocking primer or sequencing run.

I wish I was more help here.

---

Now this question, I can answer:

@Cravo_Adri

I’m having problems to define this threshold and it would be good if someone knew if there are other options to check the efficiency of the primer.

Yes. Build a true-positive control (a simple community with known composition), then sequence that true-positive control with your variense sequencing primers and blocking primers. Given that you know the true composition of the community, you can see which combination of primers works best.

True-positive controls for the win!

Colin

[Cravo_Adri (Adriana Paula Bernardo Cravo)]

Hi, @colinbrislawn !
Thanks A LOT for all the sugestions, they were helpful! Actually, I didn’t consider that the run itself could cause the problem… so I’ll do what you suggested and add samples with and without the blocking primer in the same run and if necessary build the true-positive control
thanks again!