
A
snakemake
workflow for phylogenomic analysis of genome clusters.
kizuchi
is a
snakemake
workflow for building gene trees, starting from a collection of genomes
and HMM profiles, and then analyzing the phylogenetic histories of alleles
within genome-level ANI clusters. The aim of this workflow is to automate and
document gene tree analysis in a reproducible way, with the ultimate objective
of tracing recombination across timescale domains.
-
gene prediction using
prodigal-gv
-
gene annotation using
hmmer
-
ortholog scoring
-
amino acid alignment using
clustalo
-
alignment trimming using
trimal
-
tree inference using
fasttree
-
genome clustring using
fastANI
-
phylogenetic correlations using
SuchTree
Code Snippets
85 86 87 88 89 | shell : 'rm -rf aggregated_genomes proteins genes \ hmm_hits hmmdb preani ani ani_clusters \ scored_proteins cluster_data clusters \ statistics benchmarks logs' |
99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 | run : with open( log[0], 'w' ) as LOG : fasta_chunks = batched( input.fastas, size=len(input.fastas)//config['chunks'] ) if not os.path.exists( 'aggregated_genomes' ) : os.mkdir( 'aggregated_genomes' ) for n,chunk in enumerate( fasta_chunks ) : chunkpath = os.path.join( 'aggregated_genomes', 'chunk_{n}.fna'.format(n=str(n) ) ) LOG.write( 'writing {chunk}...\n'.format( chunk=chunkpath ) ) with open( chunkpath, 'w' ) as f : contigs = 0 for i,genome in enumerate( chunk ) : for j,seq in enumerate( parse( open( genome ), 'fasta' ) ) : seq.id = Path(genome).stem + '_contig_' + str(j) seq.description = '' f.write( seq.format('fasta') ) contigs = contigs + j LOG.write( ' wrote {i} genomes with {j} contigs\n'.format( i=str(i+1), j=str(contigs+1) ) ) |
131 132 | wrapper : 'file://wrappers/prodigal-gv' |
144 145 146 147 148 149 150 151 152 | run : profiles = [] for hmmfile in input.hmms : hmm = hmmreader.read_single( open( hmmfile ) ) if Path(hmmfile).stem != hmm.metadata.model_name : raise Exception( 'file name ({hmmfile}) must match model name ({modelname}).'.format( hmmfile=hmmfile, modelname=hmm.metadata.model_name ) ) profiles.append( hmm ) hmmwriter.save_many_to_file( hmms=profiles, output=output.db ) |
168 169 | wrapper : 'file://wrappers/hmmer/hmmsearch' |
184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 | run : with open( log[0], 'w' ) as LOG : prealignments = { Path(p).stem : { 'prealign' : p, 'hits' : [] } for p in output.prealign } # use trusted cutoffs if provided trusted_cutoffs = { m.metadata.model_name : m.metadata.trusted_cutoff for m in hmmreader.read_all( open( input.profile ) ) } LOG.write( 'parsed {n} hmms, found {m} trusted cutoffs\n'.format( n=str(len(trusted_cutoffs)), m=str(sum( [ bool(v) for v in trusted_cutoffs.values() ] ) ) ) ) # load hmmer hits and protein sequences for each chunk for tblout,faa in zip( input.hits, input.faa ) : hits = polars.DataFrame( [ h for h in parse_hmmsearch( tblout ) ] ) seqs = { seq.id : seq for seq in parse( open( faa ), 'fasta' ) } chunk_genomes = { name.rsplit('_',1)[0] for name in seqs.keys() } LOG.write( '{chunk} : found {h} hits for {p} proteins in {g} genomes\n'.format( chunk=Path(faa).stem, h=str(len(hits)), p=str(len(seqs)), g=str(len(chunk_genomes)) ) ) if len( hits ) == 0 : continue # for each hmm, for each genome, write the proteins # with the top hits in order of their score value for genome,hmm in product( hits['tname'].unique(), hits['qname'].unique() ) : if trusted_cutoffs[ hmm ] : top_hits = hits.filter( ( polars.col('tname') == genome ) & ( polars.col('qname') == hmm ) & ( polars.col('score') >= trusted_cutoff[hmm] ) )\ .sort( 'score', descending=True ) else : top_hits = hits.filter( ( polars.col('tname') == genome ) & ( polars.col('qname') == hmm ) & ( polars.col('eval') <= EVALUE ) )\ .sort( 'score', descending=True ) if len( top_hits ) == 0 : continue with open( prealignments[ hmm ]['prealign'], 'a' ) as f : for n,protein in enumerate( top_hits['tname'] ) : seq = deepcopy( seqs[ protein ] ) if seq.seq[-1] == '*' : seq = seq[:-1] seq.id = seq.id + '_p' + str(n) prealignments[hmm]['hits'].append( seq.id ) f.write( seq.format( 'fasta' ) ) for k,p in prealignments.items() : genomes = { h.rsplit('_',4)[0] for h in p['hits'] } LOG.write( '{hmm} : \n'.format( hmm=k ) ) LOG.write( ' genomes : {n}\n'.format( n=str(len(genomes)) ) ) LOG.write( ' hits : {n}\n'.format( n=str(len(p['hits'])) ) ) if len( p['hits'] ) == 0 : LOG.write( ' NONE\n' ) |
247 248 249 250 251 252 253 | run : with open( output.references, 'w' ) as rf : for fasta in input.references : rf.write( fasta + '\n' ) with open( output.queries, 'w' ) as qf : for fasta in input.queries : qf.write( fasta + '\n' ) |
267 268 | wrapper : 'file://wrappers/fastani' |
280 281 282 283 284 285 286 287 288 289 290 291 292 293 294 295 296 297 298 299 300 301 302 303 304 305 306 307 308 309 310 311 312 313 314 315 316 317 318 319 320 321 322 323 324 325 326 327 328 | run : header = [ 'query', 'reference', 'ANI', 'bidirectional fragment mappings', 'total query fragments' ] fastani = pandas.read_csv( input.ani, sep='\t', names=header ) # file names -> genome names fastani['query'] = [ Path(p).stem for p in fastani['query'] ] fastani['reference'] = [ Path(p).stem for p in fastani['reference'] ] # drop self-hits fastani = fastani[ fastani['query'] != fastani['reference'] ] # de-duplicate (i.e., take the lower triangle of the matrix) fastani['hash'] = fastani.apply( lambda row : hash( tuple( sorted( ( row['query'], row['reference'] ) ) ) ), axis=1 ) fastani.drop_duplicates( subset='hash', keep='first', inplace=True ) fastani.drop( ['hash'], axis=1, inplace=True ) # convert % identity to a distance measure for clustering threshold = 1.0 - float( Path( output[0] ).stem.rsplit( '_', 1 )[1] )/100 with open( output[0], 'w' ) as f : genomes = list( set( fastani['query'] ) | set( fastani['reference'] ) ) D = zeros( ( len(genomes), len(genomes) ) ) D.fill( 1.0 ) for i in range(len(genomes)) : D[i,i] = 0.0 for n,row in fastani.iterrows() : i = genomes.index( row['query'] ) j = genomes.index( row['reference'] ) D[i,j] = D[j,i] = 1.0 - row['ANI']/100 ani_clusters = defaultdict(list) for n,c in enumerate( fcluster( single( squareform(D) ), threshold, criterion='distance' ) ) : ani_clusters[c].append( genomes[n] ) for cid, cg in ani_clusters.items() : f.write( '{cid}\t{genomes}\n'.format( cid=str(cid), genomes=','.join(cg) ) ) |
344 345 346 347 348 349 350 351 352 353 354 355 356 357 358 359 360 361 362 363 364 365 366 367 368 369 370 371 372 373 374 375 376 377 378 379 380 381 382 383 384 385 386 387 388 389 390 391 392 393 394 395 396 397 398 399 400 401 402 403 404 405 406 407 408 409 410 411 412 413 414 415 416 417 418 419 420 421 422 423 424 425 426 427 428 429 430 431 432 433 434 435 436 | run : def intersect( hmms, genomes, threshold ) : C = defaultdict( set ) for hmm,genome in zip( hmms, genomes ) : C[hmm].add(genome) return sum( [ len( C[a] & C[b] ) > threshold for a,b in combinations( C.keys(), 2 ) ] ) with open( log[0], 'w' ) as LOG : LOG.write( 'loading sequences...\n' ) sequences = { Path(fasta).stem : { rec.id : rec.seq for rec in parse( open( fasta ), 'fasta' ) } for fasta in input.prealignments } LOG.write( 'Genes loaded : {n}\n'.format( n=str(len(sequences)) ) ) geneseqs = defaultdict(dict) for chunk in input.genes : for seq in parse( open( chunk ), 'fasta' ) : genome = seq.id.rsplit('_',3)[0] geneseqs[ genome ][ seq.id ] = seq LOG.write( 'Genomes loaded : {n}\n'.format( n=str(len(geneseqs)) ) ) LOG.write( 'creating gene table for genome clusters...\n' ) records = [] for ani,clusterfile in zip( params.ANIs, input.ani_clusters ) : for cid,names in [ line.split() for line in open( clusterfile ) ] : names = names.split(',') #if not len( names ) > 5 : continue for hmm in sequences.keys() : seq_ids = [ seq_id for seq_id in sequences[hmm].keys() if seq_id.rsplit('_',4)[0] in names ] for seq_id in seq_ids : records.append( { 'seq_id' : seq_id, 'genome' : seq_id.rsplit('_',4)[0], 'cid' : int( cid ), 'hmm' : hmm, 'ANI' : float( ani ), 'id' : cid + '_' + Path( clusterfile ).stem.split('_')[-1] } ) df = polars.DataFrame( records ) LOG.write( 'number of clusters by ANI...\n' ) for ani,n in df.groupby('ANI').agg( polars.col('cid').unique().count() ).sort('ANI').iter_rows() : LOG.write( ' ANI={ani} : {n} clusters\n'.format( ani=ani, n=n ) ) LOG.write( 'gene table description :\n\n' + str( df.describe() ) + '\n' ) clusters = df.groupby('id').agg( ['hmm','genome'] ).apply( lambda x : ( x[0], intersect( x[1], x[2], MIN_CLUSTER_LINKS ) ) ).filter( polars.col('column_1') > 0 )['column_0'] LOG.write( 'found {n} genome clusters with at least {T} genomes sharing at least two genes.\n'.format( n=len(clusters), T=MIN_CLUSTER_LINKS ) ) cluster_table = df.filter( polars.col('id').is_in( clusters ) ) LOG.write( 'filtered gene table description :\n\n' + str( cluster_table.describe() ) ) cluster_table.write_csv( file=output.cluster_table, separator='\t' ) # create cluster directory structure and write gene nucleotide # sequences as individual fastas per cluster LOG.write( 'found {n} clusters.\n'.format( n=len( cluster_table.groupby('id').agg('seq_id') ) ) ) if not os.path.exists( 'clusters' ) : os.mkdir( 'clusters' ) for ANI,cids in cluster_table.groupby('ANI').agg( polars.col('cid').unique() ).sort('ANI').iter_rows() : anipath = os.path.join( 'clusters', str(ANI) ) if not os.path.exists( anipath ) : os.mkdir( anipath ) for cluster_id in cids : clusterpath = os.path.join( anipath, str( cluster_id ) ) if not os.path.exists( clusterpath ) : os.mkdir( clusterpath ) genepath = os.path.join( clusterpath, 'genes' ) if not os.path.exists( genepath ) : os.mkdir( genepath ) for hmm,seq_ids,genomes in cluster_table.filter( ( polars.col('ANI') == ANI ) & ( polars.col('cid') == cluster_id ) ).groupby('hmm').agg( ['seq_id', 'genome'] ).iter_rows() : with open( os.path.join( genepath, hmm + '.fna' ), 'w' ) as f : for seq_id,genome in zip( seq_ids, genomes ) : orig_seq_id = seq_id.rsplit('_',1)[0] # remove hmmer evalue rank seq = geneseqs[genome][orig_seq_id] seq.id = seq_id f.write( seq.format( 'fasta' ) ) |
Support
- Future updates
Related Workflows





