comparative analysis of 16S rRNA sequencing data of different runs

[Harshita_Sharma]

hi everyone,
I have 16S rRNA sequencing data from 4 runs: run1: 2x150 bp paired end sequencing using illumina miseq ; run2: 2x300 bp paired end using illumina hiseqX; run3: 2x300 bp paired end using miseq; run 4: single end sequencing using ion torrent. and i want to do a comparative analysis for which i am using qiime2.
This is what i have done till now:
i downloaded fastq.gz files from ncbi using SRA accession id and run FastQC which showed me presence of Illumina universal adapter in run 1 and 2 only which i removed using cutadapt. Then, i merged the paired end reads of run 1,2 and 3 followed by quality-filter q-score method for all the samples of 4 runs. then, i performed dereplication of all of these 4 run samples which gave me feature table and feature data as output. next, i proceeded with chimera removal using both denovo and reference based approaches, for which i got 3 output files [chimeric sequences, non-chimeric sequences and stats file].
Now, my queries are:

1. if i proceed with taxonomic annotation, how do i generate feature table for the above generated non-chimeric sequences?
2. this is what the stats file generated after chimera removal looks like:
   
   
   
   How to interpret it?
3. Am i goin in right direction ?
   Someone Please help me out!!!!!!

[cherman2 (Chloe Herman)]

Hi @Harshita_Sharma,
Seems like you are doing really exciting research. Comparative analysis are tricky.

I am personally getting alittle lost in all your steps.

@Harshita_Sharma

i downloaded fastq.gz files from ncbi using SRA accession id and run FastQC which showed me presence of Illumina universal adapter in run 1 and 2 only which i removed using cutadapt. Then, i merged the paired end reads of run 1,2 and 3 followed by quality-filter q-score method for all the samples of 4 runs. then, i performed dereplication of all of these 4 run samples which gave me feature table and feature data as output. next, i proceeded with chimera removal using both denovo and reference based approaches, for which i got 3 output files [chimeric sequences, non-chimeric sequences and stats file].

Would you mind giving us a list of exact commands you ran? I think this might help our mods better understand what you are running and how to get you where you want to go!

@Harshita_Sharma

if i proceed with taxonomic annotation, how do i generate feature table for the above generated non-chimeric sequences?

I believe you need to filter your feature-table using the non-chimeric sequences like in this post: Error with vsearch uchime-denovo output - #8 by Oddant1

@Harshita_Sharma

this is what the stats file generated after chimera removal looks like:

Here is more info regarding the columns and what they mean: Galaxy |

[Harshita_Sharma]

@cherman2 Thank you for your valuable input.
This is the list of QIIME2 methods that i have used for my analysis till now:

1. fastqc
2. cutadapt trim-paired
3. vsearch join-pairs
4. quality filter q-score
5. vsearch dereplicate-sequences
6. feature-table filter-features
7. vsearch uchime-ref
8. feature-table merge-seqs
9. feature-table merge
10. feature-classifier: for this feature i am unable to use the SILVA database as reference: it always shows this error: "Plugin error from feature-classifier:

** The scikit-learn version (0.24.1) used to generate this artifact does not match the current version of scikit-learn installed (1.4.2). Please retrain your classifier for your current deployment to prevent data-corruption errors."**
What do you suggest i should do about it?
Also, using the input provided by you for feature table generation of nonchimeric sequences, i was able to execute it properly, Thank you again for that.

[colinbrislawn (Colin J Brislawn)]

Hello Harshita,

Thank you for posting your full pipeline and full error. Here is the core of the error that also tells us what to try next:

@Harshita_Sharma

** The scikit-learn version (0.24.1) used to generate this artifact does not match the current version of scikit-learn installed (1.4.2). Please retrain your classifier for your current deployment to prevent data-corruption errors."**

You can retrain as suggested, or download a copy of the SILVA database pre-trained on the correct version of scikit-learn on this page: https://resources.qiime2.org/

Let us know if you are able to find a version of silva that matches and get this command running!

P.S. Yes, you are going in the right direction!