1. Installing fasturec. 

tar xzvf fasturec.tgz

Current version is for linux only. Other operating systems will be also available.

For source code please send request to gorecki@mimuw.edu.pl.

2. Running.

Simple help: run with no options.
Input file: use -G FILENAME
Run heuristic: use -Y
Output file: <date>.<time>.fu.txt
Log: fasturec.log (args only)

2.1. Example dataset: g3simple.txt (easy to test).

Unrooted binary trees from yeast dataset (Genolevures).
The best tree for DL (duplication-loss cost) has score 64413.
The best tree for D (duplication cost) has score 15822.

- initial tree is quasi-consensus tree (default), DL cost; quasi-consesus tree is a simple tree reconstructed from cluster frequencies of the input gene trees

fasturec -G g3simple.txt -Y

- initial tree is quasi-consensus tree (default), D cost (use -L0 to set loss weight to 0)

fasturec -G g3simple.txt -Y -L0 

- initial tree is random, DL cost

fasturec -G g3simple.txt -Y -r1 

- random initial tree, D cost.

fasturec -G g3simple.txt -Y -r1 -L0

2.2. Example dataset: ac.txt (more complex than previous).

Unrooted binary gene trees from TreeFam database (AC trees, curated).

The best tree for DL (duplication-loss cost) has score 7188. 
The best tree for D (duplication cost) has score 2240.

- running from initial random tree may lead to non-optimal solution.
E.g. compare (run several times):

fasturec -G ac.txt -Y  -r1

to

fasturec -G ac.txt -Y 

- to explore more deeply add, for example, -j10 parameter:

fasturec -G ac.txt -Y  -r1 -j10

- eventually, change the explored neighbourhood size to 2 (-k2 option):

fasturec -G ac.txt -Y  -r1 -j10 -k2

3. Additional comment.

-r option can be used with larger numbers. However, I'm not fully happy with this approach.
Try (10 random trees, and with 10 repeats): 

fasturec -G ac.txt -Y -r10 -k1 -j10

4. Output.

All trees explored during fasturec are saved to a file: date.time.fu.txt (it might be unsorted). Each line of the output file contains cost and a rooted species tree.