Hi! I have some fasta files with a couple of sequences each. They are like this:
>WP_012937914.1 NADH-dependent butanol dehydrogenase [Acidaminococcus fermentans]
MLGNFTYCNPTKLYFGKDALQNLPAELAKYGKKVVLVYGSGSIKKNGIYDAVTKILAEAGKDVAEIAGVMSNPTIHKLRE
GIAIARKHGADFILAVGGGSVIDYSKALSVSVNCDEDPWEKYFVRFEEPTCETIPVGTVLTMVGTGSEMNAGAVITYPEK
KLKMGKVFADEKIMPRFSILNPVYTMTLPKYQMVAGIYDIFNHICEQYFSGTDDNTSDYLAEGLMRNVVEASRTAVKNPQ
DYEARSNLMWDATWALNTLIAKGKSTDWMVHMLGQAAGGVTNATHGMTLAAVSLPYYRFILKDGLPKFVRFAKVVWDVRP
EGKTDEEIAKEGLEKMEDWMQQLGLVLHLKDLGATEDMLDDLVNGTLILTGGYRVLTKEEIREIFRRAM
How can I extract,from this files, only the sequences that have in that first line "NADH" ? I need that first line as well as the sequence into another fasta file.
•
written
by
chups519 •
This would be a fine use case for bioawk, or pyfaidx:
$ pip install --user pyfaidx
$ faidx proteins.fa --regex "*NADH*" > NADH_proteins.fa
I just tried this out, but although the python module installs just fine, I'm getting "command not found" when I try and run the script. Maybe it's getting copied to somewhere that isn't in my path (or not getting copied anywhere)?
OK, when I try it without the --user flag, it works, so I reckon it's that.
Yes, I should have mentioned that the --user flag places everything in $HOME/.local/bin, and so that's where the faidx script would go, instead of /usr/local/bin. Glad you've figured it out!
With Biopython you can do this as follows:
from Bio import SeqIO
your_fasta_files = ['your_first_file', 'your_second_file', ..., 'your_last_file']
nadh_entries = []
for fasta_file in your_fasta_files:
entries = SeqIO.parse(fasta_file, 'fasta')
for entry in entries:
if 'NADH' in entry.description:
nadh_entries.append(entry)
SeqIO.write(nadh_entries, 'your_nadh_fasta_output_file', 'fasta')
to be on the safe side for capitalization, I would go for:
if 'nadh' in entry.descripton.lower():
Maybe this post helps
How To Extract A Sequence From A Big (6Gb) Multifasta File ?
For now the suggestions in the posto don't work, but thank you so much for the help. I'll keep trying
I know the windows solution.
a. Keep all fasta in single folder (I assume they have .fasta extension
b. Open cmd and go to that folder
c. run
copy *.fasta merged.txt
let the commands run and close cmd. rename the merged.txt file to merged.fasta
a. open file in excel
b. assuming headers are in col a and sequences in col b, type this in col c1(including "=" sign
= if(countif(b1,"*NADH*")=0,0,1)
This will give 1 if the sequence initiates with NADH. 4 represents count for search string in this case 4 as NADH makes 4 alphabets.
c. Filter col c for value 1
d. Copy the first 2 columns and make a txt file out of it. Convert this tab file to fasta using Tab to fasta sequence converter.
AAAAAAHHHHH! Not only does this not work for the OP's example (the "NADH" is not at the beginning of the definition line) but it involves way too many manual manipulations of the data. If you want to work with FASTA and answer this question on Windows, you can install Python for Windows and just run pyfaidx from the command prompt. There's no need to open your file in Excel.
Now, since I've stated my opinion, thanks for giving your answer, and it's obvious from your solution that you have something that works, and sometimes that's all that matters. However in this case there are better tools.
My apology. The markdown language is new to me. I didnt realise the asterik sign has been ommited from the final answer. I have modified equation to show the formula to detect "NADH" anywhere in sentence"
Python will be a great solution. We do it in PHP. But sometimes, learning a program is too tidious and if jobs simply can be done with office tools like excel and word, whats the harm in it!!
This is really a very bad way to solve a problem.
FABOX has an option to detect extract sequences containing spoecific headers or part of it. its slow and probably hangs for >200 sequences. Excel on 4 gb ram can very well handle a tab sequence file upto 33000 sequences averaging to 500 aa average length. A formula and filter in such case takes approx 2-5 min. I dont think, a beginner can program anything in python in this much short time. Please do consider the time in learning the software.
Log in to answer this question.
