def annotate(gene_list, db_info): # [local_db, all_species, ensembl_db_name, species] = db_info db = connect_to_mysql() cfg = ConfigurationReader() acg = AlignmentCommandGenerator() cursor = db.cursor() if verbose: print "thread %s annotating %s " % (get_thread_name(), species) if not species == 'oryctolagus_cuniculus': print 'The preferred list of species is hardcoded for the rabbit. Consider modifying.' exit(1) preferred_species = [ species, 'mus_musculus', 'rattus_norvegicus', 'homo_sapiens' ] nearest_species_list = species_sort(cursor, all_species, species) species_list = preferred_species + filter( lambda x: x not in preferred_species, nearest_species_list) inf = erropen("temp_out.fasta", "w") for gene_id in gene_list: #for gene_id in [90020]: switch_to_db(cursor, ensembl_db_name[species]) #################### # get stable id and description of this gene stable_id = gene2stable(cursor, gene_id) if not gene_list.index(gene_id) % 100: print gene_list.index(gene_id), "out of", len(gene_list) if verbose: print "=============================================" if verbose: print gene_id, stable_id #################### # find the annotation from the preferred source organism [annot_source, orthology_type, annotation, ortho_stable_ids] = find_annotation(cursor, ensembl_db_name, species_list, gene_id) if verbose: print annot_source, "**", orthology_type, '**', annotation ################### # find splices (for now find the canonical splice) switch_to_db(cursor, ensembl_db_name[species]) canonical_splice = get_canonical_transl(acg, cursor, gene_id, species) # output if orthology_type == 'self' or annotation == 'none': header = ">{0} {1}".format(stable_id, annotation) else: header = ">{0} {1} [by sim to {2}, {3}]".format( stable_id, annotation, annot_source, ortho_stable_ids) print >> inf, header print >> inf, canonical_splice cursor.close() db.close()
def main(): db = connect_to_mysql() acg = AlignmentCommandGenerator() cfg = ConfigurationReader() cursor = db.cursor() [all_species, ensembl_db_name] = get_species (cursor) # human and mouse are the only two species that have CCDs info for species in [ 'homo_sapiens', 'mus_musculus']: alt_splice_almt (cursor, cfg, acg, species, ensembl_db_name) cursor.close() db .close()
def make_alignments (species_list, db_info): [local_db, ensembl_db_name] = db_info verbose = False flank_length = 10 db = connect_to_mysql() cfg = ConfigurationReader() acg = AlignmentCommandGenerator() cursor = db.cursor() # find db ids adn common names for each species db [all_species, ensembl_db_name] = get_species (cursor) max_days = 60 for species in species_list: species_shorthand = get_species_shorthand(cursor, species) print(species, species_shorthand) directory = check_directory (cfg, species, species_shorthand, "pep") if not directory: continue removed = 0 remaining = 0 for dirname, dirnames, filenames in os.walk(directory): for filename in filenames: full_name = os.path.join(dirname, filename) time_modified = os.path.getmtime(full_name) number_of_days_since_modified = (time.time() - time_modified)/(60*60*24) if number_of_days_since_modified > max_days: #print "removing", filename, "made", number_of_days_since_modified, "ago" os.remove(full_name) else: remaining += 1 print(species, "done, removed", removed, "files, remaining", remaining)
def main(): db = connect_to_mysql() acg = AlignmentCommandGenerator() cursor = db.cursor() [all_species, ensembl_db_name] = get_species(cursor) species = 'homo_sapiens' switch_to_db(cursor, ensembl_db_name[species]) gene_list = get_gene_ids(cursor, biotype='protein_coding') for gene_id in gene_list: # find stable stable_id = gene2stable(cursor, gene_id=gene_id) canonical = get_canonical_transl(acg, cursor, gene_id, species, strip_X=False) if canonical: print stable_id, canonical cursor.close() db.close()
def find_missing_exons(human_gene_list, db_info): # [local_db, ensembl_db_name, method] = db_info db = connect_to_mysql() cfg = ConfigurationReader() acg = AlignmentCommandGenerator() cursor = db.cursor() # find db ids and common names for each species db all_species, ensembl_db_name = get_species(cursor) # minimal acceptable similarity between exons min_similarity = cfg.get_value('min_accptbl_exon_sim') switch_to_db(cursor, ensembl_db_name['homo_sapiens']) ################################################################################## # loop over human genes gene_ct = 0 found = 0 sought = 0 unsequenced = 0 #human_gene_list.reverse() for human_gene_id in human_gene_list: switch_to_db(cursor, ensembl_db_name['homo_sapiens']) # Get stable id and description of this gene -- DEBUG human_stable = gene2stable(cursor, human_gene_id) human_description = get_description(cursor, human_gene_id) if verbose: print(human_gene_id, human_stable, human_description) # progress counter gene_ct += 1 if (not gene_ct % 10): print("processed ", gene_ct, " out of ", len(human_gene_list), "genes") print("exons found: ", found, " out of ", sought, "sought") # find all human exons for this gene that we are tracking in the database human_exons = [ e for e in gene2exon_list(cursor, human_gene_id) if e.covering_exon < 0 and e.is_canonical and e.is_known ] if not human_exons: print("\t\t no exons found") continue human_exons.sort(key=lambda exon: exon.start_in_gene) for he in human_exons: he.stable_id = exon2stable(cursor, he.exon_id) ################################################################################## ################################################################################## # make 'table' of maps, which is either pointer to the map if it exists, or None map_table = {} for species in all_species: map_table[species] = {} for he in human_exons: map_table[species][he] = None ################# maps_for_exon = {} for he in human_exons: maps_for_exon[he] = get_maps(cursor, ensembl_db_name, he.exon_id, he.is_known) # exon data for m in maps_for_exon[he]: #if m.source == 'usearch': continue #if m.source == 'sw_sharp': continue #if m.source == 'sw_sharp': # print 'sw_sharp' #if m.source == 'usearch': # print 'usearch', m.similarity, m.species_2, m.exon_id_1, m.exon_id_2 if m.similarity < min_similarity: continue m_previous = map_table[m.species_2][he] if m_previous and m_previous.similarity > m.similarity: continue map_table[m.species_2][he] = m # get rid of species that do not have the gene at all for species in all_species: one_exon_found = False for he in human_exons: if map_table[species][he]: one_exon_found = True break if not one_exon_found: del map_table[species] # fill in the peptide sequence field for each human exon # get rid of exons that appear in no other species but human (?) bad_he = [] for he in human_exons: one_species_found = False he.pepseq = get_exon_pepseq(cursor, he, ensembl_db_name['homo_sapiens']) if len( he.pepseq ) < 3: # can I ever get rid of all the nonsense I find in Ensembl? bad_he.append(he) continue for species in list(map_table.keys()): if species == 'homo_sapiens': continue if map_table[species][he]: one_species_found = True break if not one_species_found: bad_he.append(he) human_exons = [he for he in human_exons if not he in bad_he] # keep track of nearest neighbors for each human exon previous = {} next = {} prev = None for he in human_exons: previous[he] = prev if prev: next[prev] = he prev = he next[he] = None # fill, starting from the species that are nearest to the human if not list(map_table.keys()): continue # whatever species_sorted_from_human = species_sort(cursor, list(map_table.keys()), species)[1:] for species in species_sorted_from_human: print(species) # see which exons have which neighbors #if verbose: print he.exon_id, species no_left = [] no_right = [] has_both_neighbors = [] one_existing_map = None for he in human_exons: m = map_table[species][he] if m and not m.warning: # the one existing map should not be a problematic one one_existing_map = m continue prev = previous[he] nxt = next[he] if prev and nxt and map_table[species][prev] and map_table[ species][nxt]: has_both_neighbors.append(he) elif not prev or not map_table[species][prev]: no_left.append(he) elif not nxt or not map_table[species][nxt]: no_right.append(he) if not one_existing_map: continue # this shouldn't happen if not has_both_neighbors and not no_left and not no_right: continue # what is the gene that we are talking about? exon_id = one_existing_map.exon_id_2 is_known = one_existing_map.exon_known_2 gene_id = exon_id2gene_id(cursor, ensembl_db_name[species], exon_id, is_known) # is it mitochondrial? mitochondrial = is_mitochondrial(cursor, gene_id, ensembl_db_name[species]) # where is the gene origin (position on the sequence) gene_coords = get_gene_coordinates(cursor, gene_id, ensembl_db_name[species]) if not gene_coords: continue [gene_seq_region_id, gene_start, gene_end, gene_strand] = gene_coords # fill in exons that have both neighbors: # human exon functions as a coordinate here for he in has_both_neighbors: # get template (known exon from the nearest species) template_info = get_template(cursor, ensembl_db_name, map_table, species, he) if not template_info: continue # previous_ and next_seq_region are of the type Seq_Region defined on the top of the file # get previous region prev_seq_region = get_neighboring_region( cursor, ensembl_db_name, map_table, species, gene_coords, he, previous[he]) if not prev_seq_region: continue # get following region next_seq_region = get_neighboring_region( cursor, ensembl_db_name, map_table, species, gene_coords, he, next[he]) if not next_seq_region: continue sought += 1 reply = find_NNN(cursor, ensembl_db_name, cfg, acg, he, maps_for_exon[he], species, gene_id, gene_coords, prev_seq_region, next_seq_region, template_info, mitochondrial, method) if reply == 'NNN': unsequenced += 1 # work backwards # use the last known region on the left as the bound no_left.reverse() next_seq_region = None for he in no_left: m = map_table[species][he] # check first if we haave already looked into this, and found incomplete region #if m and m.warning: continue # get template (known exon from the nearest species) template_info = get_template(cursor, ensembl_db_name, map_table, species, he) if not template_info: continue # get following region if not next_seq_region: next_seq_region = get_neighboring_region( cursor, ensembl_db_name, map_table, species, gene_coords, he, next[he]) if not next_seq_region: continue # otherwise it is the last thing we found # the previous region is eyeballed from the next on # the previous and the next region frame the search region prev_seq_region = left_region(next_seq_region, MAX_SEARCH_LENGTH) sought += 1 reply = find_NNN(cursor, ensembl_db_name, cfg, acg, he, maps_for_exon[he], species, gene_id, gene_coords, prev_seq_region, next_seq_region, template_info, mitochondrial, method) if reply == 'NNN': unsequenced += 1 # repeat the whole procedure on the right prev_seq_region = None for he in no_right: m = map_table[species][he] # check first if we haave already looked into this, and found incomplete region #if m and m.warning: continue # get template (known exon from the nearest species) template_info = get_template(cursor, ensembl_db_name, map_table, species, he) if not template_info: continue # get following region if not prev_seq_region: prev_seq_region = get_neighboring_region( cursor, ensembl_db_name, map_table, species, gene_coords, he, previous[he]) if not prev_seq_region: continue # otherwise it is the last thing we found # the following region is eyeballed from the previous next_seq_region = right_region(prev_seq_region, MAX_SEARCH_LENGTH) sought += 1 reply = find_NNN(cursor, ensembl_db_name, cfg, acg, he, maps_for_exon[he], species, gene_id, gene_coords, prev_seq_region, next_seq_region, template_info, mitochondrial, method) if reply == 'NNN': unsequenced += 1 print(species, "sought", sought, " unseq", unsequenced)
def main(): verbose = True db = connect_to_mysql() acg = AlignmentCommandGenerator() cursor = db.cursor() # find db ids adn common names for each species db [all_species, ensembl_db_name] = get_species(cursor) logf = erropen("error.log", "w") if not logf: exit(1) outf = erropen("mut_significance_bg_data.txt", "w") if not outf: exit(1) switch_to_db(cursor, ensembl_db_name['homo_sapiens']) gene_ids = get_gene_ids(cursor, biotype='protein_coding', is_known=1, ref_only=True) # the categories of mutations for which we will be collecting statistics fill_category() # for each human gene #gene_ids = [10093176 ] for gene_id in gene_ids: switch_to_db(cursor, ensembl_db_name['homo_sapiens']) stable_id = gene2stable(cursor, gene_id) # find all canonical coding human exons # get_canonical_coding_exons also sorts exons by the start in the gene canonical_human_exons = get_canonical_coding_exons( cursor, gene_id, ensembl_db_name['homo_sapiens']) # bail out if there is a problem if not canonical_human_exons: continue full_reconstituted_cDNA = "" prev_codon_piece_plus_right_flank = "" for human_exon in canonical_human_exons: [exon_seq_id, pepseq, pepseq_transl_start, pepseq_transl_end, left_flank, right_flank, nucseq] = \ get_exon_seqs(cursor, human_exon.exon_id, human_exon.is_known) # add the split codon phase = get_exon_phase(cursor, human_exon.exon_id, human_exon.is_known) left_flank_plus_codon_piece = left_flank + nucseq[: pepseq_transl_start] split_codon = "" if phase > 0 and prev_codon_piece_plus_right_flank and left_flank: offset = (3 - phase) % 3 # hedge against the possibility that the translation starts # right at the start of the exon, but there is supposed to be a phase split_codon = prev_codon_piece_plus_right_flank[: phase] + left_flank_plus_codon_piece[ -offset:] full_reconstituted_cDNA += split_codon + nucseq[ pepseq_transl_start:pepseq_transl_end] prev_codon_piece_plus_right_flank = nucseq[ pepseq_transl_end:] + right_flank mitochondrial = is_mitochondrial(cursor, gene_id) if (mitochondrial): full_reconstituted_seq = Seq(full_reconstituted_cDNA).translate( table="Vertebrate Mitochondrial").tostring() else: full_reconstituted_seq = Seq( full_reconstituted_cDNA).translate().tostring() canonical = get_canonical_transl(acg, cursor, gene_id, 'homo_sapiens', strip_X=False) if canonical[ 0] == 'X': #that's some crap apparently wrong transcript is annotated as canonical print >> logf, "warning", gene_id, stable_id, get_description( cursor, gene_id) print >> logf, "the deposited canonical sequence starts with X - is there an alternative (?)" canonical = canonical[1:] if full_reconstituted_seq[-1] == '*' and canonical[-1] != '*': canonical += '*' if (len(full_reconstituted_seq) != len(canonical) or full_reconstituted_seq != canonical): if (len(canonical) - len(full_reconstituted_seq) < 3 and full_reconstituted_seq in canonical): # go with it - I do not have that much of that crap anyway print >> logf, "warning", gene_id, stable_id, get_description( cursor, gene_id) print >> logf, "missing a couple of amino acids in beginning or in the end" else: print >> logf, "error", gene_id, stable_id, get_description( cursor, gene_id) print >> logf, "error reassembling, len(full_reconstituted_seq) != len(canonical) ", len( full_reconstituted_seq), len(canonical) print >> logf, "canonical:" print >> logf, canonical print >> logf, "reconstituted:" print >> logf, full_reconstituted_seq continue # nucleotide stats count = {'A': 0, 'C': 0, 'C-CpG': 0, 'T': 0, 'G': 0, 'G-CpG': 0} is_CpG = {} for i in range(len(full_reconstituted_cDNA)): is_CpG[i] = False if full_reconstituted_cDNA[i] == 'A': count['A'] += 1 elif full_reconstituted_cDNA[i] == 'T': count['T'] += 1 elif full_reconstituted_cDNA[i] == 'C': if i + 1 < len(full_reconstituted_cDNA ) and full_reconstituted_cDNA[i + 1] == 'G': count['C-CpG'] += 1 is_CpG[i] = True else: count['C'] += 1 elif full_reconstituted_cDNA[i] == 'G': if i > 0 and full_reconstituted_cDNA[i - 1] == 'C': count['G-CpG'] += 1 is_CpG[i] = True else: count['G'] += 1 # in each category_dict (AT transt, AT transv, CG trans, CG transv, Cpg trans, cpGtransv, how many missense, # how many nonsense, how many silent possible codons = map(''.join, zip(*[iter(full_reconstituted_cDNA)] * 3)) silent = {} missense = {} nonsense = {} for cg in categories: silent[cg] = 0 missense[cg] = 0 nonsense[cg] = 0 for i in range(len(codons)): codon = codons[i] aa = full_reconstituted_seq[i] for j in range(3): nt_position = i * 3 + j nt = full_reconstituted_cDNA[nt_position] for new_nt in ['A', 'C', 'T', 'G']: if new_nt == nt: continue mutated_codon = mutate(codon, j, new_nt) if (mitochondrial): mutated_aa = Seq(mutated_codon).translate( table="Vertebrate Mitochondrial").tostring() else: mutated_aa = Seq(mutated_codon).translate().tostring() cg = category_dict[codon[j]][new_nt][is_CpG[nt_position]] if not cg or not cg in categories: print >> logf, "category problem in ", gene_id, stable_id, get_description( cursor, gene_id) print >> logf, codon, mutated_codon, j, codon[ j], new_nt, is_CpG[nt_position], cg print >> logf, i, j, nt_position, nt print >> logf, aa, mutated_aa continue if (mutated_aa == aa): silent[cg] += 1 elif (mutated_aa == "*"): nonsense[cg] += 1 else: missense[cg] += 1 print >> outf, stable_id, get_description(cursor, gene_id) print >> outf, "# CpG nucleotides (format: cdna_position|nucleotide|codon|context; )" print >> outf, "# ('context' contains one nucleotide before and one after the CpG nucleotide)" outstr = "" for i in range(len(full_reconstituted_cDNA)): if (is_CpG[i]): context = "" if i > 0: context += full_reconstituted_cDNA[i - 1] context += full_reconstituted_cDNA[i] if i < len(full_reconstituted_cDNA) - 1: context += full_reconstituted_cDNA[i + 1] outstr += "%d|%s|%s|%s;" % (i + 1, full_reconstituted_cDNA[i], codons[i / 3], context) print >> outf, outstr print >> outf, "# mutations possible (in principle)" print >> outf, "# %10s %5s %5s %5s" % ("category", "silent", "nonsense", "missense") for cg in categories: print >> outf, "%10s %5d %5d %5d" % (cg, silent[cg], nonsense[cg], missense[cg]) print >> outf, "# canonical sequence (format: <amino_acid><position_on_peptide_chain><codon>;):" outstr = "" for i in range(len(codons)): if (mitochondrial): codon_transl = Seq(codons[i]).translate( table="Vertebrate Mitochondrial").tostring() else: codon_transl = Seq(codons[i]).translate().tostring() outstr += "%s%d%s;" % (full_reconstituted_seq[i], i + 1, codons[i]) print >> outf, outstr print >> outf, stable_id, "done" logf.close()
def multiple_exon_alnmt(gene_list, db_info): print "process pid: %d, length of gene list: %d" % ( get_process_id(), len(gene_list)) [local_db, ensembl_db_name] = db_info db = connect_to_mysql() cfg = ConfigurationReader() acg = AlignmentCommandGenerator() cursor = db.cursor() # find db ids adn common names for each species db [all_species, ensembl_db_name] = get_species (cursor) species = 'homo_sapiens' switch_to_db (cursor, ensembl_db_name[species]) gene_ids = get_gene_ids (cursor, biotype='protein_coding', is_known=1) # for each human gene gene_ct = 0 tot = 0 ok = 0 no_maps = 0 no_pepseq = 0 no_orthologues = 0 min_similarity = cfg.get_value('min_accptbl_exon_sim') #gene_list.reverse() for gene_id in gene_list: start = time() gene_ct += 1 if not gene_ct%10: print gene_ct, "genes out of", len(gene_list) switch_to_db (cursor, ensembl_db_name['homo_sapiens']) print gene_ct, len(gene_ids), gene_id, gene2stable(cursor, gene_id), get_description (cursor, gene_id) human_exons = filter (lambda e: e.is_known==1 and e.is_coding and e.covering_exon<0, gene2exon_list(cursor, gene_id)) human_exons.sort(key=lambda exon: exon.start_in_gene) ################################################################## for human_exon in human_exons: tot += 1 # find all orthologous exons the human exon maps to maps = get_maps(cursor, ensembl_db_name, human_exon.exon_id, human_exon.is_known) if verbose: print "\texon no.", tot, " id", human_exon.exon_id, if not maps: print " no maps" print human_exon print if not maps: no_maps += 1 continue # human sequence to fasta: seqname = "{0}:{1}:{2}".format('homo_sapiens', human_exon.exon_id, human_exon.is_known) switch_to_db (cursor, ensembl_db_name['homo_sapiens']) [exon_seq_id, pepseq, pepseq_transl_start, pepseq_transl_end, left_flank, right_flank, dna_seq] = get_exon_seqs (cursor, human_exon.exon_id, human_exon.is_known) if (not pepseq): if verbose and human_exon.is_coding and human_exon.covering_exon <0: # this should be a master exon print "no pep seq for", human_exon.exon_id, "coding ", human_exon.is_coding, print "canonical: ", human_exon.is_canonical print "length of dna ", len(dna_seq) no_pepseq += 1 continue # collect seq from all maps, and output them in fasta format hassw = False headers = [] sequences = {} exons_per_species = {} for map in maps: switch_to_db (cursor, ensembl_db_name[map.species_2]) if map.similarity < min_similarity: continue exon = map2exon(cursor, ensembl_db_name, map) pepseq = get_exon_pepseq (cursor,exon) if (not pepseq): continue if map.source == 'sw_sharp': exon_known_code = 2 hassw = True elif map.source == 'usearch': exon_known_code = 3 hassw = True else: exon_known_code = map.exon_known_2 seqname = "{0}:{1}:{2}".format(map.species_2, map.exon_id_2, exon_known_code) headers.append(seqname) sequences[seqname] = pepseq # for split exon concatenation (see below) if not map.species_2 in exons_per_species.keys(): exons_per_species[map.species_2] = [] exons_per_species[map.species_2].append ([ map.exon_id_2, exon_known_code]); if (len(headers) <=1 ): if verbose: print "single species in the alignment" no_orthologues += 1 continue # concatenate exons from the same gene - the alignment program might go wrong otherwise concatenated = concatenate_exons (cursor, ensembl_db_name, sequences, exons_per_species) fasta_fnm = "{0}/{1}.fa".format( cfg.dir_path['scratch'], human_exon.exon_id) output_fasta (fasta_fnm, sequences.keys(), sequences) # align afa_fnm = "{0}/{1}.afa".format( cfg.dir_path['scratch'], human_exon.exon_id) mafftcmd = acg.generate_mafft_command (fasta_fnm, afa_fnm) ret = commands.getoutput(mafftcmd) if (verbose): print 'almt to', afa_fnm # read in the alignment inf = erropen(afa_fnm, "r") aligned_seqs = {} for record in SeqIO.parse(inf, "fasta"): aligned_seqs[record.id] = str(record.seq) inf.close() # split back the concatenated exons if concatenated: split_concatenated_exons (aligned_seqs, concatenated) human_seq_seen = False for seq_name, sequence in aligned_seqs.iteritems(): # if this is one of the concatenated seqs, split them back to two ### store the alignment as bitstring # Generate the bitmap bs = Bits(bin='0b' + re.sub("[^0]","1", sequence.replace('-','0'))) # The returned value of tobytes() will be padded at the end # with between zero and seven 0 bits to make it byte aligned. # I will end up with something that looks like extra alignment gaps, that I'll have to return msa_bitmap = bs.tobytes() # Retrieve information on the cognate cognate_species, cognate_exon_id, cognate_exon_known = seq_name.split(':') if cognate_exon_known == '2': source = 'sw_sharp' elif cognate_exon_known == '3': source = 'usearch' else: source = 'ensembl' if (cognate_species == 'homo_sapiens'): human_seq_seen = True cognate_genome_db_id = species2genome_db_id(cursor, cognate_species) # moves the cursor switch_to_db(cursor, ensembl_db_name['homo_sapiens']) # so move it back to h**o sapiens # Write the bitmap to the database #if (cognate_species == 'homo_sapiens'): if verbose: # and (source=='sw_sharp' or source=='usearch'): print "storing" print human_exon.exon_id, human_exon.is_known print cognate_species, cognate_genome_db_id, cognate_exon_id, cognate_exon_known, source print sequence if not msa_bitmap: print "no msa_bitmap" continue store_or_update(cursor, "exon_map", {"cognate_genome_db_id":cognate_genome_db_id, "cognate_exon_id":cognate_exon_id ,"cognate_exon_known" :cognate_exon_known, "source": source, "exon_id" :human_exon.exon_id, "exon_known":human_exon.is_known}, {"msa_bitstring":MySQLdb.escape_string(msa_bitmap)}) ok += 1 commands.getoutput("rm "+afa_fnm+" "+fasta_fnm) if verbose: print " time: %8.3f\n" % (time()-start); print "tot: ", tot, "ok: ", ok print "no maps ", no_pepseq print "no pepseq ", no_pepseq print "no orthologues ", no_orthologues print
def multiple_exon_alnmt(species_list, db_info): [local_db, ensembl_db_name] = db_info verbose = False db = connect_to_mysql() cfg = ConfigurationReader() acg = AlignmentCommandGenerator() cursor = db.cursor() for species in species_list: print print "############################" print species switch_to_db (cursor, ensembl_db_name[species]) gene_ids = get_gene_ids (cursor, biotype='protein_coding') #gene_ids = get_theme_ids(cursor, cfg, 'wnt_pathway') if not gene_ids: print "no gene_ids" continue gene_ct = 0 tot = 0 ok = 0 no_maps = 0 no_pepseq = 0 no_paralogues = 0 for gene_id in gene_ids: if verbose: start = time() gene_ct += 1 if not gene_ct%100: print species, gene_ct, "genes out of", len(gene_ids) if verbose: print print gene_id, gene2stable(cursor, gene_id), get_description (cursor, gene_id) # get the paralogues - only the representative for the family will have this paralogues = get_paras (cursor, gene_id) if not paralogues: if verbose: print "\t not a template or no paralogues" continue if verbose: print "paralogues: ", paralogues # get _all_ exons template_exons = gene2exon_list(cursor, gene_id) if (not template_exons): if verbose: print 'no exons for ', gene_id continue # find all template exons we are tracking in the database for template_exon in template_exons: if verbose: print template_exon.exon_id maps = get_maps(cursor, ensembl_db_name, template_exon.exon_id, template_exon.is_known, species=species, table='para_exon_map') if not maps: no_maps += 1 continue # output to fasta: seqname = "{0}:{1}:{2}".format('template', template_exon.exon_id, template_exon.is_known) exon_seqs_info = get_exon_seqs (cursor, template_exon.exon_id, template_exon.is_known) if not exon_seqs_info: continue [exon_seq_id, pepseq, pepseq_transl_start, pepseq_transl_end, left_flank, right_flank, dna_seq] = exon_seqs_info if (not pepseq): if ( template_exon.is_coding and template_exon.covering_exon <0): # this should be a master exon print "no pep seq for", template_exon.exon_id, "coding ", template_exon.is_coding, print "canonical: ", template_exon.is_canonical print "length of dna ", len(dna_seq) no_pepseq += 1 continue tot += 1 sequences = {seqname:pepseq} headers = [seqname] for map in maps: exon = map2exon(cursor, ensembl_db_name, map, paralogue=True) pepseq = get_exon_pepseq (cursor,exon) if (not pepseq): continue seqname = "{0}:{1}:{2}".format('para', map.exon_id_2, map.exon_known_2) headers.append(seqname) sequences[seqname] = pepseq fasta_fnm = "{0}/{1}_{2}_{3}.fa".format( cfg.dir_path['scratch'], species, template_exon.exon_id, template_exon.is_known) output_fasta (fasta_fnm, headers, sequences) if (len(headers) <=1 ): print "single species in the alignment (?)" no_paralogues += 1 continue # align afa_fnm = "{0}/{1}_{2}_{3}.afa".format( cfg.dir_path['scratch'], species, template_exon.exon_id, template_exon.is_known) mafftcmd = acg.generate_mafft_command (fasta_fnm, afa_fnm) ret = commands.getoutput(mafftcmd) # read in the alignment inf = erropen(afa_fnm, "r") if not inf: print gene_id continue template_seq_seen = False for record in SeqIO.parse(inf, "fasta"): ### store the alignment as bitstring # Generate the bitmap bs = Bits(bin='0b' + re.sub("[^0]","1", str(record.seq).replace('-','0'))) msa_bitmap = bs.tobytes() # Retrieve information on the cognate label, cognate_exon_id, cognate_exon_known = record.id.split(':') if (label == 'template'): template_seq_seen = True # Write the bitmap to the database #print "updating: ", template_exon.exon_id store_or_update(cursor, "para_exon_map", {"cognate_exon_id" :cognate_exon_id, "cognate_exon_known" :cognate_exon_known, "exon_id" :template_exon.exon_id, "exon_known" :template_exon.is_known}, {"msa_bitstring":MySQLdb.escape_string(msa_bitmap)}) inf.close() ok += 1 commands.getoutput("rm "+afa_fnm+" "+fasta_fnm) if verbose: print " time: %8.3f\n" % (time()-start); outstr = species + " done \n" outstr += "tot: %d ok: %d \n" % (tot, ok) outstr += "no maps %d \n" % no_pepseq outstr += "no pepseq %d \n" % no_pepseq outstr += "no paralogues %d \n" % no_paralogues outstr += "\n" print outstr
def exon_cleanup(gene_list, db_info): [local_db, ensembl_db_name] = db_info db = connect_to_mysql() cfg = ConfigurationReader() acg = AlignmentCommandGenerator() cursor = db.cursor() # find db ids and common names for each species db all_species, ensembl_db_name = get_species(cursor) mammals = [ 'ailuropoda_melanoleuca', 'bos_taurus', 'callithrix_jacchus', 'canis_familiaris', 'cavia_porcellus', 'choloepus_hoffmanni', 'dasypus_novemcinctus', 'dipodomys_ordii', 'echinops_telfairi', 'equus_caballus', 'erinaceus_europaeus', 'felis_catus', 'gorilla_gorilla', 'ictidomys_tridecemlineatus', 'loxodonta_africana', 'macaca_mulatta', 'macropus_eugenii', 'microcebus_murinus', 'monodelphis_domestica', 'mus_musculus', 'mustela_putorius_furo', 'myotis_lucifugus', 'nomascus_leucogenys', 'ochotona_princeps', 'ornithorhynchus_anatinus', 'oryctolagus_cuniculus', 'otolemur_garnettii', 'pan_troglodytes', 'papio_anubis', 'pongo_abelii', 'procavia_capensis', 'pteropus_vampyrus', 'rattus_norvegicus', 'sarcophilus_harrisii', 'sorex_araneus', 'sus_scrofa', 'tarsius_syrichta', 'tupaia_belangeri', 'tursiops_truncatus', 'vicugna_pacos' ] tot = 0 tot_ok = 0 for human_gene_id in gene_list: switch_to_db(cursor, ensembl_db_name['homo_sapiens']) stable_id = gene2stable(cursor, human_gene_id) description = get_description(cursor, human_gene_id) mitochondrial = is_mitochondrial(cursor, human_gene_id) #print "#############################################" #print human_gene_id, stable_id, get_description (cursor, human_gene_id) human_exons = get_ok_human_exons(cursor, ensembl_db_name, human_gene_id) for human_exon in human_exons: [ exon_seq_id, human_protein_seq, pepseq_transl_start, pepseq_transl_end, left_flank, right_flank, dna_seq ] = get_exon_seqs(cursor, human_exon.exon_id, 1, ensembl_db_name['homo_sapiens']) human_exon_phase = get_exon_phase(cursor, human_exon.exon_id, 1) first_exon = (human_exons.index(human_exon) == 0) for species in mammals: # maxentscan does not work for fish for table in ['sw_exon', 'usearch_exon']: switch_to_db(cursor, ensembl_db_name[species]) qry = "select * from %s where maps_to_human_exon_id = %d " % ( table, human_exon.exon_id) novel_exons = search_db(cursor, qry) if not novel_exons: #print "human_exon: ", human_exon.exon_id, "no", table, "for", species continue ct = 0 ok = 0 for novel_exon in novel_exons: print "%s: novel exon found in table %s, mapping to human exon %s" % \ (species, table, exon2stable (cursor, human_exon.exon_id, ensembl_db_name['homo_sapiens']) ) ct += 1 has_stop = False has_NNN = False [ novel_exon_id, gene_id, start_in_gene, end_in_gene, maps_to_human_exon_id, exon_seq_id, template_exon_seq_id, template_species, strand, phase, end_phase, has_NNN, has_stop, has_3p_ss, has_5p_ss ] = novel_exon tot += 1 exon_seqs = get_exon_seq_by_db_id( cursor, exon_seq_id, ensembl_db_name[species]) if not exon_seqs: print "exon seqs not found" continue [ exon_seq_id, protein_seq, pepseq_transl_start, pepseq_transl_end, left_flank, right_flank, dna_seq ] = exon_seqs len_ok = (pepseq_transl_end - pepseq_transl_start) == len(dna_seq) if not len_ok: # if it is not the case, then make it be so left_flank += dna_seq[:pepseq_transl_start] right_flank = dna_seq[ pepseq_transl_end:] + right_flank dna_seq = dna_seq[ pepseq_transl_start:pepseq_transl_end] pepseq_transl_start = 0 pepseq_transl_end = len(dna_seq) phase_ok = (len(dna_seq) % 3 == 0) if not phase_ok: phase = len(dna_seq) % 3 cds = dna_seq[phase:] pepseq_corrected = Seq(cds).translate().tostring() if pepseq_corrected == protein_seq: left_flank += dna_seq[:phase] dna_seq = dna_seq[phase:] else: cds = dna_seq[:-phase] pepseq_corrected = Seq( cds).translate().tostring() if pepseq_corrected == protein_seq: right_flank += dna_seq[ -phase:] + right_flank dna_seq = dna_seq[:-phase] else: print "no match ..." continue # don't want to shut-off the pipeline here pepseq_transl_start = 0 pepseq_transl_end = len(dna_seq) # retrieve the template template_db_id = species2genome_db_id( cursor, template_species) [templ_exon_seq_id, templ_protein_seq, templ_pepseq_transl_start, templ_pepseq_transl_end, templ_left_flank, templ_right_flank, templ_dna_seq] \ = get_exon_seq_by_db_id (cursor, template_exon_seq_id, ensembl_db_name[template_species]) correction = 0 phase = 0 end_phase = 0 # if this is the first exon, check if we are starting from methionine if first_exon: [left_flank_ok, correction, phase] = \ check_translation_start (mitochondrial, left_flank, dna_seq, templ_dna_seq, templ_protein_seq) # see if the left splice site is ok else: [left_flank_ok, correction, phase, max_score] = \ check_left_flank (acg, left_flank, dna_seq, templ_dna_seq) ######################## # # see if the right splice site is ok [right_flank_ok, end_correction, end_phase, end_max_score] = \ check_right_flank(acg, right_flank, dna_seq, templ_dna_seq) pepseq_corrected = "" new_left_flank = "" new_right_flank = "" new_dna_seq = "" if left_flank_ok: offset = (3 - phase) % 3 if correction: if correction > 0: new_dna_seq = dna_seq[correction:] new_left_flank = left_flank + dna_seq[: correction] else: # correction is negative, therefore left_flank[correction:] is the tail of left_flank new_dna_seq = left_flank[ correction:] + dna_seq new_left_flank = left_flank[:correction] else: new_left_flank = left_flank pepseq_transl_start = offset else: new_left_flank = left_flank if right_flank_ok: if not new_dna_seq: new_dna_seq = dna_seq if end_correction: if end_correction < 0: new_right_flank = new_dna_seq[ end_correction:] + right_flank new_dna_seq = new_dna_seq[:end_correction] else: # correction is negative, therefore right_flank[correction:] is the tail of right_flank new_right_flank = right_flank[ end_correction:] new_dna_seq += right_flank[:end_correction] else: new_right_flank = right_flank pepseq_transl_end = len(new_dna_seq) pepseq_transl_end -= end_phase else: new_right_flank = right_flank # if only one flank is ok, use that side to decide if there is a phase on the other if left_flank_ok and not right_flank_ok: end_phase = (pepseq_transl_end - pepseq_transl_start) % 3 pepseq_transl_end -= end_phase if right_flank_ok and not left_flank_ok: phase = (pepseq_transl_end - pepseq_transl_start) % 3 pepseq_transl_start += phase # check that the lengths match has_stop = None if new_dna_seq: len_old = len(left_flank + dna_seq + right_flank) len_new = len(new_left_flank + new_dna_seq + new_right_flank) if not len_old == len_new: print len_old, len_new print correction, end_correction print map(len, [left_flank, dna_seq, right_flank]) print map(len, [ new_left_flank, new_dna_seq, new_right_flank ]) continue cds = new_dna_seq[ pepseq_transl_start:pepseq_transl_end] if mitochondrial: pepseq_corrected = Seq(cds).translate( table="Vertebrate Mitochondrial").tostring( ) else: pepseq_corrected = Seq( cds).translate().tostring() if '*' in pepseq_corrected: has_stop = 1 else: has_stop = 0 if has_stop and not '*' in protein_seq: continue # abort, abort if True: print "#############################################" print human_gene_id, stable_id, "exo no:", human_exons.index( human_exon), " ", description print species, table print "\t template", template_exon_seq_id, template_species, template_db_id print "\t template left flank", templ_left_flank, templ_dna_seq[ 0:3] print "\t left flank", left_flank, dna_seq[ 0:3] print "\t ", left_flank_ok, correction, phase, if not first_exon: print max_score else: print print "\t template right flank", templ_dna_seq[ -3:], templ_right_flank print "\t right flank", dna_seq[ -3:], right_flank print "\t ", right_flank_ok, end_correction, end_phase, end_max_score print "\t human", human_protein_seq, human_exon.exon_id, human_exon_phase print "\t template", templ_protein_seq print "\t deposited", protein_seq if pepseq_corrected: print "\t corrected", pepseq_corrected if new_dna_seq: if (pepseq_transl_end - pepseq_transl_start) % 3: print "length not divisible by 3 " print pepseq_transl_start, pepseq_transl_end print phase, end_phase print len(new_dna_seq) print "%%%%% " continue else: new_dna_seq = dna_seq ######################################################### # 18_find_exons is sometimes messing up the coordinates # I do not know why ret = check_coordinates_in_the_gene( cursor, cfg, acg, ensembl_db_name, species, novel_exon, new_dna_seq) if not ret: print "\t coordinate check failed" continue [start_in_gene_corrected, end_in_gene_corrected] = ret ######################################################### # update the *_exon and exon_seq tables accordingly switch_to_db(cursor, ensembl_db_name[species]) qry = "update %s set " % table set_fields = "" if not start_in_gene_corrected == start_in_gene: if set_fields: set_fields += ", " set_fields += " start_in_gene = %d " % start_in_gene_corrected if not end_in_gene_corrected == end_in_gene: if set_fields: set_fields += ", " set_fields += " end_in_gene = %d " % end_in_gene_corrected if not has_stop is None: if set_fields: set_fields += ", " set_fields += " has_stop = %d" % has_stop if left_flank_ok: if set_fields: set_fields += ", " set_fields += " phase = %d, " % phase if first_exon: set_fields += " has_3p_ss = '%s' " % ( "first exon; starts with M") else: set_fields += " has_3p_ss = '%s' " % ( "me_score=" + str(max_score)) if right_flank_ok: if set_fields: set_fields += ", " set_fields += " end_phase = %d, " % end_phase set_fields += " has_5p_ss = '%s' " % ( "me_score=" + str(end_max_score)) qry += set_fields + " where exon_id=%d" % novel_exon_id if set_fields: search_db(cursor, qry) # update exon sequence if pepseq_corrected: # we might have changed our mind as to what is the cDNA seq, and what is flanking qry = "update exon_seq set " qry += " protein_seq = '%s', " % pepseq_corrected qry += " dna_seq = '%s', " % new_dna_seq qry += " left_flank = '%s', " % new_left_flank qry += " right_flank = '%s', " % new_right_flank qry += " pepseq_transl_start = %d, " % pepseq_transl_start qry += " pepseq_transl_end = %d " % pepseq_transl_end table_id = 2 if table == 'novel_exon' else 3 qry += " where exon_id=%d and is_known=%d" % ( novel_exon_id, table_id) search_db(cursor, qry) # gene2exon --> have to go back to 07_gene2exon for that tot_ok += 1 print "gene list done" cursor.close() db.close()
def main(): db = connect_to_mysql() acg = AlignmentCommandGenerator() cursor = db.cursor() [all_species, ensembl_db_name] = get_species(cursor) if len(sys.argv) > 1: species_list = sys.argv[1:] else: species_list = all_species ############################ for species in species_list: print print "############################" print species switch_to_db(cursor, ensembl_db_name[species]) if (species == 'homo_sapiens'): gene_ids = get_gene_ids(cursor, biotype='protein_coding', is_known=1) else: gene_ids = get_gene_ids(cursor, biotype='protein_coding') ct = 0 tot = 0 for tot in range(1000): #for gene_id in gene_ids: #tot += 1 gene_id = choice(gene_ids) # find all canonical coding exons associated with the gene id exons = get_canonical_coding_exons(cursor, gene_id) if (not exons): ct += 1 print gene_id, gene2stable( cursor, gene_id=gene_id), " no exons found ", ct, tot if not tot % 100: print species, tot, ct # add up the coding length of the canonical exons exons.sort(key=lambda exon: exon.start_in_gene) inside_the_coding_range = False start_properly_marked = False length = 0 for exon in exons: if not exon.canon_transl_start is None: start_properly_marked = True # if it is not propermy marked, we'll never start reading inside_the_coding_range = True length -= exon.canon_transl_start - 1 if not exon.canon_transl_end is None: inside_the_coding_range = False length += exon.canon_transl_end if inside_the_coding_range: length += exon.end_in_gene - exon.start_in_gene + 1 # take that all exons are coding full length if there is no start and end annotation # (this I believe is the case for predicted transcripts) if not start_properly_marked: length = 0 for exon in exons: length += exon.end_in_gene - exon.start_in_gene + 1 if (not length): print gene2stable( cursor, gene_id=gene_id), " no exons marked as canonical" continue # what is the length of the canonical transcript according to Ensembl canonical_translation = get_canonical_transl(acg, cursor, gene_id, species, strip_X=False) if (not canonical_translation): print "no canonical transl found for ", gene_id continue if (abs(length / 3 - len(canonical_translation)) > 3): ct += 1 print gene_id, gene2stable(cursor, gene_id), get_description( cursor, gene_id) print "(length of all exons)/3 ", length / 3, print " does not match reported canonical transl len ", len( canonical_translation) if False: # print out all exons print "exons:" inspect(exons) print print 'canonical sequence' print re.sub( "(.{50})", "\\1\n", canonical_translation ) # print canonical sequence with \n stuck in every 50 positions print # print out exons more carefully filtered to belong to the canonical version of the translation print get_translated_region_talkative(cursor, gene_id, species) all_exons = gene2exon_list(cursor, gene_id) print "all exons:" inspect(all_exons) print compare_seqs(canonical_translation, translated_seq, verbose=False) exit(1) print species, "checked a sample of ", tot + 1, "genes; problematic:", ct cursor.close() db.close() # # print 'Note: some problems could not have be resolved up to this point,' # print 'becasue we have not really looged at the exons seqs yet.' # print 'For example, for MP furo the, start fo the cannonical translation' # print 'is sometimes given in the middle of NNNNN region.' # return True