Snakefile.consensu

import glob
import numpy as np

def get_replicates (wildcards):
	files=list()
	for file in glob.glob("output/RAxML_bestTree.ASC_GTRGAMMA_felsenstein.{genome}.*".format(genome=wildcards.genome)):
		files.append(file)
	return(files)


workingdir="/nfs/pgsb/projects/evograph/analysis/paper_triae1/manuscript_whealbi/after_NatureGen_review/unimputed/RRHS_RAxML/"
replicates=1000
REPL=list(range(1, replicates+1))

GENOMES=np.array(["B","A","D"])
GENOMES=np.repeat(GENOMES,4,axis=0)

METHODS=np.array(["MajorityRuleExtendedConsensusTree",
				  "MajorityRuleConsensusTree",
				  "StrictConsensusTree",
				  "Threshold-75-ConsensusTree"])
METHODS=np.tile(METHODS,3)

METHODS_SHORT= np.array(["MRE","MR","STRICT","T_75"])
METHODS_SHORT=np.tile(METHODS_SHORT,3)

rule all:
	input:
		expand("RAxML_{method}.ASC_GTRGAMMA_felsenstein.{genome}.RRHS_consensus.{method_short}",zip,genome=GENOMES, method=METHODS,method_short=METHODS_SHORT)

rule compute_consensus:
	input:
		"RAxML_bestTree.ASC_GTRGAMMA_felsenstein.{genome}.1000.boot"
	output:
		"RAxML_{method}.ASC_GTRGAMMA_felsenstein.{genome}.RRHS_consensus.{method_short}"
	shell:
		"/home/pgsb/daniel.lang/software/standard-RAxML/raxmlHPC-PTHREADS-SSE3 -m ASC_GTRGAMMA -J {wildcards.method_short} -z {input} -n ASC_GTRGAMMA_felsenstein.{wildcards.genome}.RRHS_consensus.{wildcards.method_short} -T 20"

rule concat_RRHS_replicates:
	input:
		get_replicates
	output:
		"RAxML_bestTree.ASC_GTRGAMMA_felsenstein.{genome}.1000.boot"
	shell:
		"cat {input} > {output}"