We will be hosting mothur and R workshops throughout 2018. Learn more.
The pre.cluster command implements a pseudo-single linkage algorithm with the goal of removing sequences that are likely due to pyrosequencing errors. A version of this algorithm was developed by Sue Huse (paper here). The basic idea is that abundant sequences are more likely to generate erroneous sequences than rare sequences. With that in mind, the algorithm proceeds by ranking sequences in order of their abundance. Then we walk through the list of sequences looking for rarer sequences that are within some threshold of the original sequence. Those that are within the threshold are merged with the larger sequence. The original Huse method performs this task on a distance matrix, whereas we do it based on the original sequences. The advantage of our approach is that the algorithm works on aligned sequences instead of a distance matrix. This is advantageous because by pre-clustering you remove a large number of sequences making the distance calculation much faster.
The pre.cluster command expects a fasta-formatted file and a names or count file and that the sequences are in the same order in both files. Both of these files can be generated by the unique.seqs command. For example, if you are following along with the Sogin data analysis example and have aligned, filtered, and unique'd your sequences, then enter the following to perform the pre.clustering command:
mothur > unique.seqs(fasta=sogin.unique.filter.fasta, name=sogin.names) mothur > pre.cluster(fasta=sogin.unique.filter.unique.fasta, name=sogin.unique.filter.names)
mothur > unique.seqs(fasta=sogin.unique.filter.fasta, count=sogin.count_table) mothur > pre.cluster(fasta=sogin.unique.filter.unique.fasta, count=sogin.unique.filter.count_table)
Will result in the following output:
0 21821 86 100 20286 1621 200 19824 2083 ... 21700 16380 5527 21800 16377 5530 21900 16376 5531
Total number of sequences before precluster was 21907. pre.cluster removed 5531 sequences.
This output indicates, by column, the number of sequences processed, the number of sequences that will be found in the final dataset, and the number of sequences that have been clustered away. This should accelerate as the function runs. In this example, this step merged 5,531 sequences with other sequences, leaving you with a set of 16,376 sequences to work with. As an additional benefit to removing potentially erroneous sequences, the reduced dataset will run about 1.8 times faster through dist.seqs than the original and should cluster much faster as well. Two files are created - a *.precluster.fasta and a *.precluster.names file containing the new sequence and names file or *.precluster.count_table for further processing.
If you provide a groupfile or your count file contains group information, mothur will pre.cluster sample by sample.
mothur > pre.cluster(fasta=stool.trim.unique.good.filter.unique.fasta, name=stool.trim.unique.good.filter.names, group=stool.good.groups)
mothur > pre.cluster(fasta=stool.trim.unique.good.filter.unique.fasta, count=stool.trim.unique.good.filter.count_table)
The screen output will look like:
Processing group F11Fcsw: 0 355 5 100 292 68 200 284 76 300 281 79 360 281 79 Total number of sequences before pre.cluster was 360. pre.cluster removed 79 sequences. ...
By default the pre.cluster command will look for sequences that are within 1 mismatch of the sequence being considered. With the diffs option you can change this threshold. For example:
mothur > pre.cluster(fasta=sogin.unique.filter.unique.fasta, name=sogin.unique.filter.names, diffs=2) 0 21777 130 100 19165 2742 200 18419 3488 ... 21700 13273 8634 21800 13270 8637 21900 13267 8640
Total number of sequences before precluster was 21907. pre.cluster removed 8640 sequences.
When using unaligned sequences, the pre.cluster command allows you to select between three alignment methods - blastn, gotoh, and needleman - needleman is the default setting:
mothur > pre.cluster(fasta=sogin.unique.filter.unique.fasta, name=sogin.unique.filter.names, diffs=2, align=needleman)
The needleman algorithm penalizes the same amount for opening and extending a gap. Alternatively, you could use the gotoh algorithm, which charges a different penalty for opening (default=-2) and extending (default=-1) gaps:
mothur > pre.cluster(fasta=sogin.unique.filter.unique.fasta, name=sogin.unique.filter.names, diffs=2, align=gotoh)
Our experience has shown that the added parameters in the gotoh algorithm do not improve the pairwise alignment and increases the time required for the alignment. Finally, blastn can be used as a heuristic approach to the gotoh alignment:
mothur > pre.cluster(fasta=sogin.unique.filter.unique.fasta, name=sogin.unique.filter.names, diffs=2, align=blast)
In our implementation, blast is the slowest option of the three and also generates the worst alignments. The quality suffers particularly because it generates a local alignment (needleman and gotoh are global) and will truncate the alignment if the similarity drops below a threshold.
match, mismatch, gapopen, and gapextend
If you are using unaligned sequences, in the pairwise alignment portion of the aligning procedure, the default reward for a match is +1 and the penalties for a mismatch, opening and extending a gap are -1, -2, and -1. Our experience has shown that these produce the best alignments for 16S rRNA gene sequences. You are encouraged to play around with these to suit your own purposes as shown below:
mothur > pre.cluster(fasta=sogin.unique.filter.unique.fasta, name=sogin.unique.filter.names, diffs=2, match=1, mismatch=-3)
mothur > pre.cluster(fasta=sogin.unique.filter.unique.fasta, name=sogin.unique.filter.names, diffs=2, gapopen=-5)
Keep in mind that if you are using the align=blast option, blast will limit the combinations of match, mismatch, gapopen, and gapextend that you can use. Hopefully, we've scared you off of using blast at all so that this won't be an issue.
The topdown parameter allows you to specify whether to cluster from largest abundance to smallest or smallest to largest. Default=T, meaning largest to smallest.
As shown above, pre.cluster expects you to provide a name file so that it can acquire the abundance information from each sequence. If you do not provide the name file the command will automatically run your data through unique.seqs to generate to get the information it needs.
Something to keep in mind is that when you set the number of mismatches to 2, you are allowing that the maximum difference between sequences within a cluster to be 4 (2 from the dominant sequence in one direction, and 2 in any other direction). This difference of 4 bases, could compromise your ability to distinguish signal from noise. For example, with this Sogin datasets, the sequences are ~60 bp V6 pyrotags. A difference of 4 bases is 6.7%! Alternatively, when using diffs=1, the difference of 2 bases is 3.3%. The assumption of the algorithm is that these mismatches are noise; however, it doesn't make sense to then analyze your data at 3% by either level of diffs. Remember that this method does not actually remove the noise, it just clusters sequences that are likely to be noisy. To remove the noise you would need to use a program like Chris Quince's PyroNoise or AmpliconNoise. Considering using PyroNoise/AmpliconNoise may not be practical for many people, the pre.cluster option may be your best bet.
- 1.22.0 Added group option.
- 1.23.0 Added processors option for by group processing.
- 1.23.0 Added map file to output.
- 1.28.0 Added count option
- 1.30.0 Added topdown parameter
- 1.33.0 Improved work balance load between processors.
- 1.36.0 Added cluster method for unaligned sequences. Added align, mismatch, match, gapopen, gapextend parameters.
- 1.39.0 Major speed improvement for pre.cluster command with aligned files.
- 1.39.0 Major memory requirement reduction for pre.cluster command with by group option.
- 1.39.1 Pre.cluster(without groups) not clustering properly