How to generate a consensus sequence

Demultiplexing

As sequencers create large numbers of reads, the library being sequenced can consist of multiple samples that are pooled together for sequencing. To determine which reads have come from which sample in the pool, unique barcode sequences (one barcode for each sample being sequenced) are added during the library preparation step. When the library is sequenced, each read will contain (at its ends) the unique barcodes that are used to label the samples.

Demultiplexing is the process of sorting sequencing reads based on the unique barcode sequences into different files, with each file representing one of the samples being sequenced. Both Illumina (through BCL convert or BaseSpace) and Oxford Nanopore Technology (ONT) (through guppy or EPI2ME) platforms can demultiplex automatically. However, you must provide the software with a list of sample names and the corresponding barcode sequences used for each sample. It is important to note that demultiplexing is not 100% accurate. For example, sometimes multiple sequencing errors occur within the barcode of a read leading to a read being undetermined or assigned to the wrong sample (misassigned).

Data download

Although there are some cloud based systems (such as BaseSpace and EPI2ME) which offer a number of options to analyse your data, it is typical to transfer sequence reads onto your local computer or server for analysis. Read files can be large and are therefore usually compressed to reduce data storage needs and speed up data transfer processes. If you are using BaseSpace or EPI2ME then you can simply download your FASTQ reads through their corresponding web interfaces onto your local machine.

Alternatively, you can utilise file transfer software that can transfer data from one machine (e.g. a sequencer) to another machine (e.g. a computer) over the internet. There are several free tools that can be used to do this, such as Cyberduck and FileZilla, which have easy to use graphical interfaces. It is standard to use the Secure File Transfer Protocol (SFTP) option within these tools as this encrypts the data while it is being transferred. Security is especially important when dealing with human sequence data.

It is important to note that your raw sequence data is very valuable, and it is therefore very important to keep adequate backups of your data in case of any data loss disasters.

Data quality control

We now have our sequence reads demultiplexed into different sample FASTQ files. The first step in any analysis is quality control (garbage in = garbage out) which typically involves evaluating the number and quality of the reads. Common QC steps are removing any leftover adapter sequences from the ends of reads, removing poor quality reads, and trimming poor quality ends from reads (primarily for Illumina reads, as sequenced base quality typically deteriorates along the read length). The overall aim is to remove potentially erroneous reads (or erroneous parts of reads) from the data, which will improve downstream analyses such as reference alignments, consensus and variant calling.

FASTQC is one of the most popular tools used for visual inspection of read quality, as is Trim Galore! for adapter and quality trimming of Illumina reads; although there are many alternatives such as PRINSEQ, Trimmomatic, and Sickle, to name a few.

For ONT data, tools for inspection and summary statistics of reads include PycoQC, NanoStat, NanoR, and NanoQ.