Output: files and settings

Output prefix

All files generated by RAxML-NG will be given a name with the following pattern: $PREFIX.raxml.$FILETYPE. Here, $PREFIX is the name of the analysis and $FILETYPE designates type of the output (see table below). By default, $PREFIX equals to the alignment file name: e.g., if you invoke RAxML-NG as follows:

./raxml-ng --msa test.fa --model GTR+G

the best-scoring ML tree will be saved to test.fa.bestTree (in the same directory). Alternatively, you can specify a different output prefix with the --prefix option. Prefix can either be a simple name, or contain both analysis name and path to the output files:

--prefix myRun (output files will be placed into the current directory)

--prefix myFolder/myRun (relative path + name)

--prefix /home/alex/myFolder/myRun (absolute path + name)

Output files

Depending on the type of analysis, RAxML-NG will generate a subset of the following files:

Filename	Description
$PREFIX.raxml.ancestralProbs	Ancestral state reconstruction (ASR): marginal state probabilities
$PREFIX.raxml.ancestralStates	ASR: Reconstructed ancestral sequences
$PREFIX.raxml.ancestralTree	ASR: Node-labeled tree
$PREFIX.raxml.bestTree	Best-scoring ML tree (strictly bifurcating)
$PREFIX.raxml.bestTreeCollapsed	Multifurcating version of the best-scoring ML tree. As of v1.0.0, only near-zero branches (<= blmin) are collapsed. In future releases, this can be extended to include e.g. branches with low support etc.
$PREFIX.raxml.bestPartitionTrees	Best-scoring per-partition ML trees (in --brlen unlinked mode)
$PREFIX.raxml.bestModel	Optimized model parameters for the best-scoring ML tree
$PREFIX.raxml.bootstraps	Trees inferred for each bootstrap replicate
$PREFIX.raxml.bootstrapMSA.<REP>.phy	Bootstrap replicate alignments
$PREFIX.raxml.consensusTree	Consensus tree file
$PREFIX.raxml.log	Screen log
$PREFIX.raxml.mlTrees	ML trees for each starting tree
$PREFIX.raxml.rba	Compressed MSA in binary format (RAxML Binary Alignment)
$PREFIX.raxml.rfDistances	Pair-wise RF distances among input trees
$PREFIX.raxml.startTree	Starting tree(s) used for ML inference
$PREFIX.raxml.support	Best-scoring ML tree with bootstrap support values
$PREFIX.raxml.terrace	All trees residing on a terrace (compressed Newick)
$PREFIX.raxml.terraceNewick	All trees residing on a terrace (multi-line standard Newick)

NOTE: You can suppress generation of all output files with the --nofiles switch.

In addition, following intermediate files will be created while RAxML-NG is running:

Filename	Description
$PREFIX.raxml.ckp	Checkpoint file
$PREFIX.raxml.lastTree.TMP	Best-scoring ML tree found so far
$PREFIX.raxml.mlTrees.TMP	ML trees inferred so far
$PREFIX.raxml.bootstraps.TMP	Bootstrap replicate trees inferred so far

NOTE: You can suppress generation of all intermediate files (excl. checkpoints) with the --nofiles interim switch. Typically, the overhead of writing the intermediate files in negligible, but under certain conditions (lots of small trees, many coarse-parallel searches) disabling it can accelerate inference by a few percent.

Re-running the analysis

By default, RAxML-NG checks whether files with the same PREFIX already exist. If the previous run was interrupted and a checkpoint file is available, RAxML-NG will resume execution from the checkpoint. If the previous run finished successfully and result files were created, RAxML-NG will exit with an error. This is a safety measure to prevent accidental overwriting of existing results. If desired, this behaviour can be disabled: just add --redo switch, and RAxML-NG will re-run the analysis from scratch, ignoring an existing checkpoint and overwriting the result files.

Log verbosity

You can control the amount of information printed to the terminal/log file with the --log option. Following log levels are supported (from least to most verbose):

Log level	Description
--log ERROR	Silent (only error messages)
--log WARNING	+ Print warnings
--log RESULT	+ Print results, e.g. final tree log-likelihood
--log INFO	+ Print additional info, e.g. alignment dimensions (default)
--log PROGRESS	+ Print search progress messages
--log VERBOSE	+ Print even more details. e.g. evolutionary model parameters
--log DEBUG	+ Print diagnostic messages