You could use the all barcode gap discovery tool (see the website http://wwwabi.snv.jussieu.fr/public/abgd/) they have an online tool. Larger datasets can be handled here as well. For the method see (Mol Ecol. 2012 Apr;21(8):1864-77. doi: 10.1111/j.1365-294X.2011.05239.x. Epub 2011 Aug 29.

ABGD, Automatic Barcode Gap Discovery for primary species delimitation.

Puillandre N1, Lambert A, Brouillet S, Achaz G). I think it is downloadable via researchgate or other media. note this method is tree independent! but widely used and can help seperate OTU´s with a distance based criterium.

For a NJ reconstruction the MEGA software comes to mind. This also comes with good manuals, but NJ tree is easy to make (http://www.megasoftware.net/). 

THe parsimony and Bayesian methods are very computaionally demanding. But Mr Bayes and PAUP* are software that is used for these methods. For a phylogenetic method the GMYC method can be used (see: Delimiting Species Using Single-locus Data and the Generalized Mixed Yule Coalescent (GMYC) Approach: A Revised Method and Evaluation on Simulated Datasets by Tomochika Fujisawa and Timothy G. Barraclough Syst Biol (2013) doi: 10.1093/sysbio/syt033)

Good luck

Thank you,  @Floris Breman, I ve seen ABGD, this tool is for distance based analysis. While, MEGA provides basic level analysis till construction of tree. Even Mr Bayes and PAUP are used till tree construction. Pls have a look over the image of table no. 4, their is comparison of no of spp/no of seq, which includes successful, failed, misidentified and ambigious/singletons. Please suggest software or method to get similar type of results.

Dear Rahul

I saw the figure, the software used whas not directly clear, of course I suggested ones that are rather common and would repeat part of what you showed. I am afraid I do not see exactly what you want further. Trees can be reconstructed with most of these softwares and you can use that as result then (if they do not take your4000 sequences contact the people who made it, sometimes they have versions online or for larger datasets, but I have put on occasion >800 sequences through MEGA without problems), judging from the results not much is wrong. Results for each method are comparable and that suggest the patterns is real. Further resulst are dependent on threshold and number of sequences per species (with BM and BCM approach) and how you define a succes. (number of species identified or number of sequences identified. If not still not clear you may mail with more explanation and or results.

Hope this helps

Sincerely

Thank you for your reply @ Floris Breman, I really appereciate your contribution in developing Adhoc package, which is a milestone in DNA barcoding and species identification. Actually I am analyzing sequences from BOLD and Genbank for assessment of DNA Barcodes and species identificaiton. For this I ve used two methods one distance based and other tree based.

Distance based method:

Adhoc was used as it employs K2P (distance based) method for species discrimination. In this I ve imported DNA barcode library in Adhoc using 'checkDNAAbcd' function, while ‘adhocTHR’ function was used to find Redflagged, BCM and BM. This processess was repeated till redflagged and FP (Ambigious) species were discarded completely. Finally I got true speces, which confirms species identification based on distance method.

Tree based method:

In this we should discriminate species according to formation of clusters

that is number of clusters in numerical form (for example Catla: 2 clustersof  ambigious sequences, 2 clusters of ambigious species, 4 clusters of true species), also, different methods like NJ, Mr. Bayes and Parsimony could be used in this cluster analysis.

Or else please suggest possible method to discriminate species on tree based method, keep posting.

Thank you!

I go with the answers of  Bremen, try Mixed Yule Coalescent (GMYC) through R programming and sample size may nt be a problem. best of luck.

Thank you Pavithra for your suggestion!

I believe the only accurate method would be alignment-based phylogenetic analysis based on likelihood (e.g. RAxML on a public server such as CIPRES for such a large dataset), combined with a GMYC approach. Anything else, including and particular clustering and NJ methods, have been shown to be inaccurate, sometimes substantially so. Also, if you actually want to TEST barcoding accuracy, you need a testable expectation. For example, you have a set of barcoding sequences for KNOWN species (or at least species-level clades), based on a reconstructed phylogeny. Then you randomly select a sequence and blast it against the entire dataset and see how often the correct species is blasted. That is kind of trivial but that would be the only way of testing. If you have a set of unidentified barcoding sequences and just want to see how many distinct species you have in there, that would not be a barcoding test but an assessment of species delimitation. A very quick approach to sorting out species would also be using MAFFT with the sorting option, which will align the sequences and also sort them based on similarity. It is not 100% accurate but usually, especially if all sequences are complete, does a great job in outlining possible species-level clades even before running a tree. And 4000 sequences can be easily done in MAFFT.

Thank you Robert Lücking for your suggestion, I will go with it and return to the post.

Go for the advise of Robert Lüking especially if you have computer clusters etc at your disposal.  MAFFT is a really nice allignment tool that can indeed sort the sequences based on similarity. The online version is also really fast (in my experience anyway). Further sequences from BOLD/GENBANK should be checked and their ID´s should be ok, particularly those in BOLD...at least so the theory goes. This has been demonstraterd time and time again not to be true. I would double check with specialists or authors (who submitted the sequences) in case of problems with particular taxa . I would also check the literature for known problems.

Just discarding false positives, finally, seems to be a bit strange. There can be perfectly good reasons for a sequence to end up in the FP (the samples for which the INTERspecific K2P distances are lower than the threshold) category. Hybridisation, incomplete lineage sorting, wrong ID´s, sample/sequence mixups. The problem is of course what the real cause is. The first two are biological processes and should be considered/discussed as they influence any succes rate you for DNA barcoding. Even phylogeny based methods will run into problems with these samples.

Be also very carefull what you call a ´true species´ many reviewers will demand a good explanation and good arguments for the use of such a term. What are the underlying concepts you use (phylogenetic, morphological, distance, biological concept etc etc...) I assume you take the ID´s of those who submitted the sequences yes? Many studies took biological/morphological ID´s as a starting point and then discussed DNA barcodes (but not all).

Finally DNA barcoding does not have to be perfect in identifying 100% of the species (and indeed rarely is) all the time, in my opinion , so long as you can identify most of your samples and can discuss the rest, either with an expert of the group or simply discuss the pattern you observe in your paper so that others can continue the study at some other time or with another method. DNA barcoding is about identifying of course so even if only one base consistently separates taxa it isa realiable indicator so I think distance methods, even though they are very imperfect still have a use especially in closely related taxa. Distance methods may even outperforme phylogeny/tree based methods here.

Sincerely

Floris

Thank you, Floris Breman!