It took a while to chase down where this error is coming from. It turns out to be in the graph package:

> showMethods("acc", classes="graph", includeDefs=TRUE)
Function: acc (package graph)
object="graph", index="character"
function (object, index) 
{
    nN <- numNodes(object)
    nNames <- nodes(object)
    nIndex <- length(index)
    whN <- match(index, nNames)
    if anyis.na(whN))) 
        stop("unmatched node provided")
    rval <- vector("list", length = nIndex)
    names(rval) <- nNames[whN]
    for (i in seq_len(nIndex)) {
        marked <- rep(0, nN)
        distv <- rep(0, nN)
        names(distv) <- nNames
        distx <- 1
        names(marked) <- nNames
        nmkd <- 0
        marked[index[i]] <- 1
        done <- FALSE
        while (!done) {
            minds <- nNames[marked == 1]
            for (node in minds) {
                avec <- adj(object, node)[[1]]
                avec <- avec[marked[avec] == 0]
                marked[avec] <- 1
                distv[avec] <- distx
            }
            marked[minds] <- 2
            distx <- distx + 1
            newmk <- sum(marked == 1)
            if (newmk == 0) 
                done <- TRUE
        }
        marked[index[i]] <- 0
        rval[[i]] <- distv[marked == 2]
    }
    return(rval)
}

My best guess is that your sample names are somehow not matching up with the expected values. If you could share your data so I have a reproducible example, I could look into it further. If that's not an option, you could try seeting options(error=recover), which will help you figure out what is going on at the point in the code when that error is thrown. The above method is called from within pcairPartition (which is called from within pcair), so it might be easier to diagnose the problem if you run pcairPartition on its own.