I am a researcher working in an academic setup and I have the chance to interact with the graduate students. The trend I am seeing is that several new graduate students are drifting more towards python. I can think of two reasons. First, as Oscar pointed out, Python has a relatively lower learning curve compared to Perl. The second reason which I don't very well understand is that more professors are insisting their students to address their assignments in Python. The mandatory indentation keeping the code clean might be one big plus. Personally, I see both Perl and Python equally capable in handling my needs. However, I am more of an old school guy, so don't want to stray away too far away from perl. But at the same time, I want to keep up with the trend, so I make it a point to try a few problems in Python. For sure, Python is catching up, but I don't think it will replace perl any time soon.

I think Perl will keep its role for a while due to its easy way and power managing regular expressions. Python has maybe a more low learning curve but I think is not as powerful as Perl.

I am a researcher working in an academic setup and I have the chance to interact with the graduate students. The trend I am seeing is that several new graduate students are drifting more towards python. I can think of two reasons. First, as Oscar pointed out, Python has a relatively lower learning curve compared to Perl. The second reason which I don't very well understand is that more professors are insisting their students to address their assignments in Python. The mandatory indentation keeping the code clean might be one big plus. Personally, I see both Perl and Python equally capable in handling my needs. However, I am more of an old school guy, so don't want to stray away too far away from perl. But at the same time, I want to keep up with the trend, so I make it a point to try a few problems in Python. For sure, Python is catching up, but I don't think it will replace perl any time soon.

Power of regular expressions is same in Perl and Python. Differences are minor in this area. Libraries are quite rich on both sides. I cannot judge now if Perl caught up in numerical analysis, but 7-8 years ago it was behind in it, and ahead in many other areas.

Perl has a problem. It kind of encourages writing a short program which next month is faster to write again than understand what it's doing. One needs discipline to write readable code. Python, on the other hand, encourages readability. For me and my friends, ease of comprehending code written in Python was the reason to abandon Perl. Both own code and somebody's else.

Perl has been used traditionally for the bioinformatics community, but it has some advantages from a bigger community, new (relatively) modules such as DBIx::Class, facilitate the interaction with a database without long SQL statements, Moose push the Object Oriented (OO) programing to a more stable level, Catalyst do a cleaner web programing or it has a good modules to create Test. Right now Bioperl is been feeding from the "new" Perl modules. CPAN do your life more easy, and I think that is one of the missing pieces in Python. Basic Perl is easy to learn, nevertheless learn how you can use these other modules can take a while. There are a lot of examples of Perl bioinformatics tools widely used by the scientific community such as GBrowse or Maker.

For other hand, Python is a cleaner language, it is more difficult to write "bad code", something common in Perl. It is language that was design with the OO in mind. I don't think that it is more difficult to learn than Perl, and a long term, a more readable language is also a more easy to maintain.

There are also other languages to consider such as Ruby, with the advantage that is even newer than Python and it uses Gems, so it is easy to install new modules, or Java, perhaps more difficult to learn, but with much better performance than Perl, Python or Ruby.

Perhaps we can get one idea of the use of these languages in the scientific community looking into number of citations of Bioperl (Stajich JE et al 2002), Biopython (Cock PJA et al 2009), Bioruby (Goto N. et al 2010) and Biojava (Prlic A et al. 2012) in 2012.

Bioperl 128, Biopython 75, Bioruby 20 and Biojava 2.

So I agree with Dr. Podicheti about Python is catching up Perl, but for now is one of the most used languages in bioinformatics.

One of the advantages that Perl has is that its highly optimized for string-based operations, so working with DNA sequences will be extremely fast in Perl. Unfortunately, Perl has a steep learning curve and has low readability, but you can get around that by taking the time to write readable code that you'll be able to reference back to later. Perl has lots of tempting "one-liners", that is, reducing entire functions worth of code down to a single line. I try to avoid that practice as much as possible, after all, the compiler will optimize the code anyways. In other words, take the time to write code in a readable format and let the compiler worry about optimization. Most of the standard functions in Perl are already optimized and compiled in C, so the chances that a standard user will be able to write a faster version of say Sort, Push, Shift, etc. are not very high.

Another way to make Perl more readable is to highly comment your code as you go along. Here's an example:

sub add_two_numbers {

# This function adds two numbers and returns their sum.

my $number1 = shift; # Input the first number

my $number2 = shift; # Input the second number

my $new_number = $number1 + $number2; # Adds two numbers

return $new_number;

}

Also, use descriptive variables as I've used in the above example. So in summary, sacrifice whitespace for readability, comment as you go, use descriptive variables, and avoid "one-liners".

All that being said, if you're not working on string-based operations and readability isn't a concern, you can write C functions directly into Perl using the Inline C module from CPAN. This will give you the numerical power of a compiled C program more closely similar to that of Python's numerical capabilities, although you have to dabble in C, which has its own challenges. However, until Python's bioinformatics library can match that of BioPerl, I would suggest sticking to Perl.

Big thanks to everyone for providing answers.

Personally I am drifting away from Perl. I started out with Perl, I still own Perl for its support for me. But I find it is very difficult to transit from small programs to big programs in Perl. This can be done via OOP, but jumping from procedural Perl to OOP Perl was a pain for me. It may not be a problem to someone with good background in OOP, I started out with procedural, and Perl is not a good OOP starter. I can learn OOP with other languages, i.e. Python or Java and then come back to Perl? Not quite so, it is better learn Python and Java and keep using them if possible.

In addition, it is Big Data age, and to my experience, Perl is not doing much to it. Last time I tried with parallel and distributed Perl modules, they are very difficult to start up. I felt one has to be a hardcore Perl programmer and quite a geek in computer science to make use of them. At least that is what the authors of those modules are :-) On the other hand, it is much easier for me to understand and write parallel programs in Python or Java. The cause might be traced back to Perl's OOP and readability.

For smooth object orientated programming, check out Moose for Perl. http://moose.perl.org/ . It will save you a lot of the headache that Perl has for setting up objects. As for parallel computing, try Perl threads http://perldoc.perl.org/threads.html , but remember only certain solutions to certain problems will benefit from actual parallel computing versus running multiple instances of a program. Unfortunately, I don't think BioPerl is thread-safe (from my experience).

Python is easier to learn and encourages code that is readable. There are more ways to code something in Perl, but I don't see that as an advantage--rather, it provides more opportunity for writing cryptic code that has less reusability.

Regular expression in Perl is more handy in extracting the information from various sequence formats (Like GenBank, EMBL..etc). Especially Bioperl will ease lots of our efforts by using their predefined functions. But I don't see many people using Bioperl for their work. The problem in Bioperl might be object oriented concepts incorporated in it. Generally people feel OOPS is difficult to understand.

I am a researcher in computer science with specialization in bioinformatics. For doing my research work and dealing with gene sequences and protein sequences in my work I found Perl programming much easier with its easy way of string handling. Use of regular expressions are also much useful. Though I have not used Python but have seen the language, don't know how much it can help in biological analysis.

Perl is helpful to do bioinformatics prog.. it includes bioperl to do such operations.

@John Boyd: creating fortran or c based modules for python is equally easy. Several possibilities exist such as swig, cython, f2py, etc.

Seems like I should look at Python - but what is really important is the question one is wanting to answer and getting that right.

I started with Perl but fell for the intuitive way of writing code in Python and haven't regretted the switch since. However, I think that both languages will be around for the foreseeable future. The fasta and genbank parsers of python are great for instance, but I think the phylo module in bioperl is better than the one in biopython (at least for the moment).

One place I find Perl useful is the command line, because of its stream processing and in-place editing capabilities. For me it is easier to use than awk and sed. To do things like taking the first, fourth and fifth column of a GFF file and writing them out in a chromosome:start-end format instead:

$ cat myfile.gff | perl -p -i -e "substitution/regexp/here/"

Maybe Python and other languages have these possibilities too, but the terse code of Perl is not only more tolerable on the command line for me, it is outright desirable there.

I used to teach a "bioinformatics tool building course" that relied on Perl, CPAN, and the Ensembl APIs. It seems to me that today's graduate students like Python more and tend to turn to it instead of Perl. If I teach the course next fall, I'd probably include a big dose of Python.

No programming language is completely suited for a field (and the same goes for operating systems, their features and their default configurations). You should evaluate each task to see if the programming language meets your needs in terms of efficiency and coverage. While one may have a preference of one language over another because of ease of use or nice features, the evaluation should be made based on the needs of the project.

As a biochemist/molecular biologist, I have found the learning curve to be much easier with python. Many of the previous entries list the various reasons for this. In particular, I am very appreciative of the helpful community at large, many of which do not participate in the academic/science realm, but are willing to assist me in solving large data generated problems that seem to be the trend with my own datasets. That being said, I started with perl and still utilize simple one-liner scripts to perform specific tasks in my workflow.