Are sequencing reads written differently when inserts and indexes are sequenced seperately?

I was going over different sequencing library protocols and noticed that the adapter sequence can vary between protocols. In some adapter sequences the index sits between the primer binding site and the insert. In this case during sequencing the index will be sequenced and then the insert that comes right after it. In other cases, the adapter sequence is built so that the index is outside the primer binding site, i.e. index - primer binding site - insert (see two leftmost parts of the attached figure). Here the insert is sequenced using the Read1 primer and afterwards the index is sequenced in a separate reaction using the 17 index primer.

In this case are the index reads just added on to the ends of the insert reads in the FastQ files? Is there a way to tell from the FastQ files where the index was positioned within the adapter?

• Software FastQC reports overrepresented kmers per position in read, perhaps that could help you figure out the order? Also, note that some sequencing providers are doing some of the trimming before they hand users the data. Nov 2, 2020 at 10:46
• I do think this is a bioinformatics question. Ideally you could ignore this as an implementation detail that doesn't concern bioinformatics, but in reality I've had to stare at diagrams exactly like this when trying to troubleshoot strange illumina runs and interpret changes in base quality, for example. I'd like to see a clear answer to this myself. Nov 2, 2020 at 16:35

The order of sequencing in a dual index flow cell is:

2. Index 1 (I7)
3. Index 2 (I5)
To answer the question, the index sequences are not included in the fastq file output from Illumina, or other machines. Illumina sequencers just use the index sequence to demultiplex the complex mixture of molecules in the sequencing pool. If a small "dot" on the glass sequencing slide has some sequence, but the indices don't match what the technician specified for a library, then that read is not included in the fastq output of that library since it could be contamination of some sort.
The sequences in the fasta file theoretically all should be the insert from the library that you prepared, but sometimes the molecule being sequenced is short, and there is read-through of the adapter sequence. This is the only circumstance under which the i7 or i5 index sequences might be in the normal fastq output sequence line. This is not ideal, and this is why people perform adapter trimming.