Help with CD-hit commands for DNA sequences

Question

[frcamacho]

Hi,

I want to use cd-hit to stringently cluster and remove redundant DNA sequences (~14,500 sequences). I was doing a blastn all vs all and filtering for 98% qcovsHSP and 98% percent identity. Then running a script to find out of all the matches to keep the longer sequences. However, I found cd-hit and this allows me to do the same , but also keeps track of the clusters for me. I was going through the commands and found some that would do what I have been doing 1) removing 98% query coverage and 98% percent identity 2) keep the longer sequence in a match.

Here is what I got to try to replicate a blastn all vs all: (Please correct me if I am wrong!)

cdhit -i input.fa -o output.fa -n 11 -g 1 -G 0 -aL .98

-n word size -g accurate mode -G local sequence identity -aL # of bases in longer sequence in alignment / longer sequence length

However, I can't seem to find an argument for percent identity. I want 98% of the bases to match correctly in alignment. Any help will be appreciated!

[h.mon]

If you are dealing with DNA, you probably want to use cdhit-est

[frcamacho]

I thought cdhit -est were not good for large sequences. My max is 71KB large. Is cdhit-est ok?

Answer 1

[abascalfederico]

Hi, In the version I have the maximum % identity is controlled through "-c"

    -c  sequence identity threshold, default 0.9
this is the default cd-hit's "global sequence identity" calculated as:
number of identical amino acids in alignment
divided by the full length of the shorter sequence

[frcamacho]

Which version are you running? I am running 4.6 (built on Jul 29 2016)

[abascalfederico]

Mine is version 4.5.4, but I've checked more recent versions still use -c