The Biostrings package has many facilities for pattern/string matching. The Multiple Alignments and Pairwise Sequence Alignments vignettes would be useful to read through as a starting point.

As for getting sequences straight from UniProt, the answer to that is also yes. The UniProt website has a section on accessing its resources programmatically, which you should read through.

There are many ways to interact with a webservice in R. I recently used their webservice to do ID cross-referencing, and utilized the httr package to do so.

Since you want to fetch FASTA/peptide sequences, your queries won't look like this, but the code tidbit below that maps a series of entrez ids to UniProt accession numbers should help to get you started:

library(httr)
entrez <- c('51692', '1478', '26986')
params <- list(
               from='P_ENTREZGENEID',
               to='ACC', 
               format='tab', 
               query=paste(entrez, collapse=' '))
response <- POST('http://www.uniprot.org/mapping/', 
                 body=params, 
                 encode='multipart')
result <- read.table(textConnection(content(response, 'text')),
                     stringsAsFactors=FALSE,
                     header=TRUE)

and result is

    From         To
1| 51692     G5E9W3
2| 51692     Q9UKF6
3|  1478     P33240
4| 26986 A0A024R9C1
5| 26986     P11940