Question about cd-hit-otu's parameter

Question

[qhsh9713]

I have a 2 questions about cd-hit-otu's parameter.

The cd-hit-otu have six steps.

1. Raw read filtering and trimming
2. Clustering of duplicates
3. Chimeric reads detection
4. Read recruitment allowing one error
5. Removing noise
6. OTU clustering

and this is the parameter for first step.

  -P 0
  -i fastq-file-of-raw-reads
  -o output-fasta-file
  -t trim_cutoff, default 1.0 (means no trimming)
    if cutoff is a integer number > 1 (like 200), the program will trim sequences to this length
    if cutoff is a fraction (like 0.8), the program will keep fraction of this reads
  -p prefix-length/primers_file
    (a) if a primers_file is provided,
         read primers from this file, remove the reads don't match the primers
    (b) if a prefix-length (a digit number) is provided, default 6
         get the consensus of prefix of the all reads
         remove the reads without this prefix

specifically, I can not understand about -t and -p parameters.

Q1. what is the meaning of if cutoff is a fraction (like 0.8), the program will keep fraction of this reads?

I just understand this if parameter is less than 1(ex. 0.8), the program will trim the sequence like sequence * 0.8.

It's mean if sequence length less than sequence * 0.8, it will be removed. Is it right?

Q2. what is the meaning of prefix-length?(-p's (b))

Answer 1

[gb]

If you choose 0.8 it will trim off 20% of the read. So a read of 100 base pairs will be trimmed to 80 and a read of 200 base pairs will be trimmed to 160.

I dont know what you mean with "if the sequence length is less than sequence*0.8".

Everything depends on your goal and type of marker gene, but remember that for clustering a global alignment is made so two reads of the same origin but a different length can end up in two different clusters. That's why there is a trimming option.