Fix docs typo [skip CI]

docs/tcoffee_quickstart_regressive.rst

@@ -10,7 +10,7 @@ This document introduces the regressive mode of T-Coffee (please cite the `Natur
 
 Multiple sequence alignments (MSA) are usualy estimated by progressively aligning all the sequences, starting with the most similar. The regressive alignment is a new procedure that proceeds the other way around and starts by aligning the most distantly related sequences first. Like its progressive sibblin, the regressive algorithm starts with a guide tree. It uses this guide tree to extract the N most diverse sequences. In this first intermediate MSA, each sequence is either a leaf or the representative of a subtree. The algorithm is re-aplied recursively onto every representative sequence until all sequences have been incorporated in an internediate MSA of max size N. The final MSA is then obtained by merging all the intermediate MSAs into the final MSA. The merging is very efficient and does not require further alignment because the intermediate MSAs contain common sequences. 
 
-Aside from its improved accuracy, the main benefit of this protocol is to allow any third party multiple sequence aligner to be applied on the sub-groups (ClustalO, MAfft, etc) and any third party guide-tree protocol. The default T-Coffee supports a large number of `internal and external package <https://tcoffee.readthedocs.io/en/latest/tcoffee_main_documentation.html#internal-external-methods>`_ but if the one you need is not supported, note that it is relatively simple to create a configuration file allowing T-Coffee to interact with the package you need as docimented `here <https://tcoffee.readthedocs.io/en/latest/tcoffee_main_documentation.html#advanced-method-integration>`_. 
+Aside from its improved accuracy, the main benefit of this protocol is to allow any third party multiple sequence aligner to be applied on the sub-groups (ClustalO, MAfft, etc) and any third party guide-tree protocol. The default T-Coffee supports a large number of `internal and external package <https://tcoffee.readthedocs.io/en/latest/tcoffee_main_documentation.html#internal-external-methods>`_ but if the one you need is not supported, note that it is relatively simple to create a configuration file allowing T-Coffee to interact with the package you need as documented `here <https://tcoffee.readthedocs.io/en/latest/tcoffee_main_documentation.html#advanced-method-integration>`_. 
 
 When deployed serialy, the regressive algorithm scales linearly with the number of sequences both in time and memory. The amount of memory it needs is roughly the amount of memory required to align N sequences with the selected alignment method. Given TOT sequences, the amount of time is roughly TOT/N*(amount of time required to align N Seq with the selected method). The algorithm implementation is multi-threaded and owing to its embarrasingly parralel nature, it benefits from a very efficient speedup when deployed this way.