yet another text parser...

[Marc Hoeppner]

Hi,

I have yet another question about how to write a specific text parser in
ruby...
So, without further ado - this is what the source file looks like:

Query= gi|23510597|emb|CAD48982.1| ring-infected erythrocyte surface
antigen precursor [Plasmodium falciparum 3D7]
(1085 letters)

Database: KOG
112,920 sequences; 47,500,486 total letters

Searching..................................................done



Score
E
Sequences producing significant alignments: (bits)
Value

At2g21510 96
3e-19
At4g39150 95
1e-18
At1g76700

and so on...

What I want to do is the following:
Read the source file - and if a line starts with "Query=", strip
everything from the line but the expression "gi|xxxxx". That part was no
problem with gsub, mind you. But, now the tricky thing (or not, I
guess...).
Go from there until you find a line starting with "Sequence", skip this
line and the following and puts the third line together with the
"gi|xxxxx"
So from the above example it would look like this:

gi|23510597 At2g21510

No, ideally I wouldnt have to include this skip-lines part, but I cant
find a regexp, that lets me reliably identify the first line of the
results block (not all possible results start with At...).

How I tried to do it:

def stripname line
s = line.gsub(/Query=/, '')
u = s.gsub(/\|emb.*/, '')
end

count = 0 # initializing variables
t = nil
v = nil

ARGF.each do |l|

puts l unless count.zero?
count = [0, count-1].max

if l.match(/^Query=/)
t = stripname l
elsif l.match(/^Sequences/)
l = $1
count = 2
puts "#{t}#{l}"
else
end
end

But the output looks terrible:
gi|23510597

At2g21510
96 3e-19
gi|23510599

At5g14980
58 3e-08
gi|23510600

And no matter what I try, I cant get the gi|xxxx and the corresponding
"best hit" in the same line. Tried it with hashes, but frankly dont know
enough about those yet.
So If anyone has a helpful comment or solution, I would be extremely
grateful!

Cheers,

Marc

 

[Andreas Schwarz]

I'd throw it all into one big ugly regex:
s.match(/Query=
(.+?\|.+?)\|.+?\(bits\)\s+Value\s+(.+?)\s+/m).to_a[1..2].join(' ')
=> "gi|23510597 At2g21510"

 

[Marc Hoeppner]

I'd throw it all into one big ugly regex:
s.match(/Query=
(.+?\|.+?)\|.+?\(bits\)\s+Value\s+(.+?)\s+/m).to_a[1..2].join(' ')
=> "gi|23510597 At2g21510"

Thanks for the suggestion! However, if someone has a suggestion
regarding the following code and how to fix it, I'd be happy...its
almost working and I just need to understand why it is behaving a bit
odd. So here is the code.

def stripname line
s = line.gsub(/Query=/, '')
u = s.gsub(/\|emb.*/, '')
end


count = 0
gene = nil
store = Array.new

ARGF.each do |l|

store.push(l) unless count.zero?
count = [0, count-1].max

if l.match(/^Query=/)
gene = stripname l

elsif l.match(/^Sequences/)
count = 2
puts "#{gene.strip} #{store.last.to_s.strip}"
else

end
end



Problem:

Reads: If line is found that starts with "Query=", use the method
stripname on it and store it in the variable "gene". Go further, and if
you find a line that starts with "Sequence", use the above specified
procedure "count". Now this is the problem right now. After I wasnt able
to figure out to get the formatting right, I decided to stick to the
skip-line approach and instead of having it printed, to store it in an
array. From there I simply read the last entry.

BUT: instead of printing every stored hit to the corresponding "gene",
it shifts the whole thing 1 line. So that each "gene" is associated with
the "best hit" of the previous match to "Query=".

gi|23510597
gi|23510599 At2g21510
gi|23510600 At5g14980

Now, I could solve that easily with a capable text editor, but I think
there must be an easy solution to this...right?

Cheers,
Marc

 

[Robert Dober]

Hi,

I have yet another question about how to write a specific text parser in
ruby...
So, without further ado - this is what the source file looks like:

Query= gi|23510597|emb|CAD48982.1| ring-infected erythrocyte surface
antigen precursor [Plasmodium falciparum 3D7]
(1085 letters)

Database: KOG
112,920 sequences; 47,500,486 total letters

Searching..................................................done



Score
E
Sequences producing significant alignments: (bits)
Value

At2g21510 96
3e-19
At4g39150 95
1e-18
At1g76700

and so on...

What I want to do is the following:
Read the source file - and if a line starts with "Query=", strip
everything from the line but the expression "gi|xxxxx". That part was no
problem with gsub, mind you. But, now the tricky thing (or not, I
guess...).
Go from there until you find a line starting with "Sequence", skip this
line and the following and puts the third line together with the
"gi|xxxxx"
So from the above example it would look like this:

gi|23510597 At2g21510

No, ideally I wouldnt have to include this skip-lines part, but I cant
find a regexp, that lets me reliably identify the first line of the
results block (not all possible results start with At...).

How I tried to do it:

def stripname line
s = line.gsub(/Query=/, '')
u = s.gsub(/\|emb.*/, '')
end

count = 0 # initializing variables
t = nil
v = nil

ARGF.each do |l|

puts l unless count.zero?
count = [0, count-1].max

if l.match(/^Query=/)
t = stripname l
elsif l.match(/^Sequences/)
l = $1
count = 2
puts "#{t}#{l}"
else
end
end

But the output looks terrible:
gi|23510597

At2g21510
96 3e-19
gi|23510599

At5g14980
58 3e-08
gi|23510600

And no matter what I try, I cant get the gi|xxxx and the corresponding
"best hit" in the same line.

It is a terrible thing happens to me all the time, one tends to forget
these \n's.
Well fortunately we have #chomp, but maybe you want to use #strip
which removes trailing (and leading) WS \n included.

HTH
Robert

 

[Robert Dober]

I'd throw it all into one big ugly regex:
s.match(/Query=
(.+?\|.+?)\|.+?\(bits\)\s+Value\s+(.+?)\s+/m).to_a[1..2].join(' ')
=> "gi|23510597 At2g21510"

Thanks for the suggestion! However, if someone has a suggestion
regarding the following code and how to fix it, I'd be happy...its
almost working and I just need to understand why it is behaving a bit
odd. So here is the code.

def stripname line
s = line.gsub(/Query=/, '')
u = s.gsub(/\|emb.*/, '')
end


count = 0
gene = nil
store = Array.new

ARGF.each do |l|

store.push(l) unless count.zero?
count = [0, count-1].max

if l.match(/^Query=/)
gene = stripname l

elsif l.match(/^Sequences/)
count = 2
puts "#{gene.strip} #{store.last.to_s.strip}"
else

end
end



Problem:

Reads: If line is found that starts with "Query=", use the method
stripname on it and store it in the variable "gene". Go further, and if
you find a line that starts with "Sequence", use the above specified
procedure "count". Now this is the problem right now. After I wasnt able
to figure out to get the formatting right, I decided to stick to the
skip-line approach and instead of having it printed, to store it in an
array. From there I simply read the last entry.

BUT: instead of printing every stored hit to the corresponding "gene",
it shifts the whole thing 1 line. So that each "gene" is associated with
the "best hit" of the previous match to "Query=".

Are you pushing before or after you use the last element of the array?
But you should go back to your original idea, which works just fine,
now that you have discovered #strip, before my post

Now this is a Ruby ML, right, so maybe you would accept that I Rubyish
the code a little bit

gi = nil
ARGF.each do |line|
case line
when /Query=\s*(gi\|.*?)\|/
gi = $1
when /Sequence/
puts gi.strip << " " << (1..2).map{ ARGF.readline }.last.strip
end
end

HTH
Robert

 

[Marc Hoeppner]

Now this is a Ruby ML, right, so maybe you would accept that I Rubyish
the code a little bit

gi = nil
ARGF.each do |line|
case line
when /Query=\s*(gi\|.*?)\|/
gi = $1
when /Sequence/
puts gi.strip << " " << (1..2).map{ ARGF.readline
}.last.strip
end
end

Very nice, thank you!