def leaderboard():
alert = None
today = datetime.date.today()
if today == datetime.date(2018, 4, 24):
alert = "Remember to submit your work for peer review on LearningSpace by midnight tonight!"
elif today == datetime.date(2018, 4, 25):
alert = "Remember to complete your assigned peer reviews on LearningSpace by midnight tomorrow!"
elif today == datetime.date(2018, 4, 26):
alert = "Remember to complete your assigned peer reviews on LearningSpace by midnight tonight!"
elif today <= datetime.date(2018, 4, 30):
alert = "Remember to check MyFalmouth for the assessment deadline!"
stats = [statistics.get_stats(bot["_id"] + '+' + class_name)
for bot in db.bots.find({})
for class_name in (bot.get("class_names") or [])]
for stat in stats:
author, bot_id, class_name = stat["_id"].split('+')
stat["bot"] = db.bots.find_one({"_id": author + '+' + bot_id})
stats.sort(key=lambda s: s["elo"], reverse=True)
unready_bots = list(db.bots.find({"status": {"$ne": "ready"}}))
matches_left = db.match_queue.find({}).count()
if matches_left > 0:
num_workers = len(glob.glob("run_matches*.pid"))
time_left = matches_left * statistics.average(m["end_time"] - m["start_time"] for m in db.match_history.find({})) / num_workers
else:
time_left = None
return flask.render_template("index.html", stats=stats, unready_bots=unready_bots, time_left=time_left, alert=alert)
def bot_info(bot_id):
bot = db.bots.find_one({"_id": bot_id})
if bot is not None:
bot_stats = [statistics.get_stats(bot_id + '+' + class_name) for class_name in bot["class_names"]]
bot_stats.sort(key = lambda s: s["elo"], reverse = True)
return flask.render_template("bot_info.html", bot=bot, stats=bot_stats)
else:
return flask.render_template("error.html", message="No bot named '%s'" % bot_id)
# print(x)
if __name__ == '__main__':
train_set = [
'./datasets/j30/' + i for i in listdir('./datasets/j30')
if i != "param.txt"
]
validation_set = []
for i in range(1, 480, 10):
validation_set.append("./datasets/RG300/datasets/RG300_" + str(i) +
".rcp")
all_rg300 = ["./datasets/RG300/" + i for i in listdir('./datasets/RG300')]
test_set = [i for i in all_rg300 if i not in validation_set]
hard_starts = [101, 141, 261, 301, 421, 461]
hard_test_tmp = []
for i in hard_starts:
for j in range(i, i + 20):
hard_test_tmp.append("./datasets/RG300/datasets/RG300_" + str(j) +
".rcp")
hard_test = [i for i in hard_test_tmp if i not in validation_set]
res = statistics.get_stats(instance,
series_priority_rules,
types,
'parallel',
'forward',
use_precomputed=False,
custom_set={'RG300': hard_test},
verbose=True)
def call_genotype(sam, chromA, chromB, posA_start, posA_end, posB_start,
posB_end, db):
s_arr = np.empty((0, 12), int) # Soft clipping alignment
m_arr = np.empty((0, 4), int) # matched alignment
with open(sam, "r") as sam_in:
#===========================================================================================
# Read in sam file and check for soft clips inside the specific region.
#===========================================================================================
for line in sam_in:
breakA = 0 # Breakpoint A (start) will be calculated below
breakB = 0 # Breakpoint B (alt mapping) will be calculated below
if line[0] == "@": # Skip info lines
continue
if line[0] == '\n': # if newline in end of file, skip this
continue
else:
line = line.upper().rstrip().split("\t")
denovo_tool = line[0]
alt_chrA = str(line[2])
if '.' in alt_chrA: # if the chromosome number contain a "." , it will be invalid and we will continue with next SV
continue
contig_start = int(line[3]) # start position for contig
map_scoreA = int(line[4])
cigar = line[5]
strandA = bam_flag(line[1])
if "S" in cigar:
bad_quality = False # If there are several possible mate-mapping positions, this will be classified as bad quality and we will ignore these Breakpoints
SA = False # Second mapping position
count_split_posA, cigar_length_posA = cigar_count(
cigar, strandA)
breakA += int(contig_start) # Breakpoint A
breakA += count_split_posA
# look at mate position of split reads. Can be found at optional field starting with SA:Z
for field in line:
if field.startswith("SA:"):
split_info = field.split(":")
positions = split_info[-1]
n_position = positions.split(
";"
) # split into number of positions. If more than one alternative position, skip!
if len(n_position
) > 2: # due to one extra object; new line
bad_quality = True
break
position = n_position[0].split(",")
alt_chrB = str(position[0])
mate_pos_start = position[1]
# strand
if position[2] == "+":
strandB = 0
elif position[2] == "-":
strandB = 1
map_scoreB = position[4]
count_split_posB, cigar_length_posB = cigar_count(
position[3], strandB)
breakB += int(mate_pos_start)
breakB += count_split_posB
SA = True
if field.startswith("AS:"):
field = field.split(":")
contig_l = field[-1]
if bad_quality:
continue
if SA == False: # If the split contig have no second mapping place, continue
continue
# count number of cigars, more cigars indicates untrustworthy SV.
cigar_length = 0
cigar_length += cigar_length_posA
cigar_length += cigar_length_posB
# check if breakpoints fall inside desired region
region = False
chromA = str(chromA)
chromB = str(chromB)
if alt_chrA == chromA and alt_chrB == chromB:
if breakA >= posA_start and breakA <= posA_end and breakB >= posB_start and breakB <= posB_end:
region = True
elif breakA >= posB_start and breakA <= posB_end and breakB >= posA_start and breakB <= posA_end:
region = True
elif alt_chrA == chromB and alt_chrB == chromA:
if breakA >= posB_start and breakA <= posB_end and breakB >= posA_start and breakB <= posA_end:
region = True
if region: # If region is True, save to array
seq = line[9]
s_arr = np.append(s_arr,
np.array([[
strandA, alt_chrA, breakA,
map_scoreA, strandB, alt_chrB,
breakB, map_scoreB, cigar_length,
int(contig_l), seq, denovo_tool
]]),
axis=0)
# Matched contig
elif "M" in cigar:
count_match_pos, cigar_length_m = cigar_count(
cigar, strandA
) # count the number of base pairs that match to reference genome
match_region_end = 0
match_region_start = int(contig_start)
match_region_end += match_region_start
match_region_end += count_match_pos
m_arr = np.append(m_arr,
np.array([[
strandA, alt_chrA,
match_region_start, match_region_end
]]),
axis=0)
#========================================================================================
# Continue analysing the cpntigs, or if no soft clipped contigs were found; look at read
# coverage over region
#========================================================================================
# if no breakpoints could be found.
if len(s_arr) == 0:
print 'No SV could be found using de novo assembly. Checking if read coverage information could be used to classify DEL, DUP and genotype'
sv_type = ""
genotype2 = ""
if chromA == chromB:
pos1 = posA_start + 1000
pos2 = posB_start + 1000
statistics = get_stats(db, chromA, pos1, pos2, 'same')
# If there is no information about read coverage, skipp this variant
if len(statistics) == 0:
return s_arr, 'N/A', 'N/A', statistics
# DELETION
if statistics['RD_norm_1'] < (0.25 * statistics['RD_all']):
sv_type = "DEL"
genotype2 = "1/1"
if statistics['RD_norm_1'] >= (0.25 * statistics['RD_all']):
if statistics['RD_norm_1'] <= (0.75 * statistics['RD_all']):
sv_type = "DEL"
genotype2 = "0/1"
# DUPLICATION
if statistics['RD_norm_1'] >= (1.25 * statistics['RD_all']):
if statistics['RD_norm_1'] < (1.75 * statistics['RD_all']):
sv_type = "DUP"
genotype2 = "0/1"
if statistics['RD_norm_1'] >= (1.75 * statistics['RD_all']):
sv_type = "DUP"
genotype2 = "1/1"
if sv_type != "" and genotype2 != "":
sv_info = [chromA, pos1, pos2]
# add genotype2 to stat dictionary
statistics['genotype2'] = genotype2
return sv_info, genotype2, sv_type, statistics
else:
return s_arr, 'N/A', 'N/A', 'N/A'
else:
return s_arr, 'N/A', 'N/A', 'N/A' # returns N/A in order to continue the loop
elif len(s_arr) > 0: # if one or more breakpoint was found
# If several predicted SVs in s_arr; use the one with longest contig
if len(s_arr) > 1:
# Best breakpoint will be the one with largest mapped contig.
seq_col = s_arr[:, 9]
seq_col = seq_col.astype(np.int)
best_bp_number = np.argmax(seq_col)
best_breakpoint = s_arr[best_bp_number]
if len(s_arr) == 1:
best_breakpoint = s_arr[0]
# Check if there is a matching contig to ref that will span over the predicted breakpoint. If there is; The genotype will be
# classified as heterozygous. If we can't find any matching contig spanning the breakpoint, we classify this as homozygous
genotype1 = ""
for row in m_arr:
if row[1] == best_breakpoint[1] and best_breakpoint[2] > row[
2] and best_breakpoint[2] < row[3]:
genotype1 = "0/1"
if genotype1 == "":
genotype1 = "1/1"
if best_breakpoint[1] != best_breakpoint[
5]: # breakpoints are located on different chromosomes -> break end
sv_type = "tBND"
genotype2 = "NA"
statistics = get_stats(db, best_breakpoint[1], best_breakpoint[2],
best_breakpoint[6], best_breakpoint[5])
# If there is no information about read coverage, skipp this variant
if len(statistics) == 0:
return s_arr, 'N/A', 'N/A', statistics
print best_breakpoint, genotype1, sv_type, statistics, 'tBND'
statistics['genotype2'] = genotype2
return best_breakpoint, genotype1, sv_type, statistics
# Get statistics from SVGenT.db and read_cov.db
if best_breakpoint[1] == best_breakpoint[5]: # if same chromosome
statistics = get_stats(db, best_breakpoint[1], best_breakpoint[2],
best_breakpoint[6], 'same')
# If there is no information about read coverage, skipp this variant
if len(statistics) == 0:
return s_arr, 'N/A', 'N/A', statistics
stat_map_score = statistics['map_1']
# mappability threshold, we do not want to keep SVs who have a low mappability score = no support for SV.
if stat_map_score < 0.25:
return s_arr, 'N/A', 'N/A', statistics
# classify SV.
else:
if best_breakpoint[1] == best_breakpoint[
5]: # breakpoints are located on the same chromosome
print 'located on same chromosome'
#===========================================================
# INV
#===========================================================
if best_breakpoint[0] != best_breakpoint[
4]: # sequences are in opposite directions -> inversed
sv_type = "INV"
genotype2 = "none"
#===========================================================
# CNV
#===========================================================
else:
# DELETION
if statistics['RD_norm_1'] < (0.25 *
statistics['RD_all']):
sv_type = "DEL"
genotype2 = "1/1"
elif statistics['RD_norm_1'] >= (
0.25 * statistics['RD_all']
) and statistics['RD_norm_1'] <= (
0.75 * statistics['RD_all']):
sv_type = "DEL"
genotype2 = "0/1"
# DUPLICATION
elif statistics['RD_norm_1'] >= (1.25 *
statistics['RD_all']):
if statistics['RD_norm_1'] < (
1.75 * statistics['RD_all']):
sv_type = "DUP"
genotype2 = "0/1"
if statistics['RD_norm_1'] >= (
1.75 * statistics['RD_all']):
sv_type = "DUP"
genotype2 = "1/1"
else:
sv_type = "BND"
genotype2 = "none"
# add genotype2 to stat dictionary
statistics['genotype2'] = genotype2
return best_breakpoint, genotype1, sv_type, statistics
def show_stats():
stats, date_list, daily_requests, sound_stats = get_stats() #gets the values from statistics.py
return render_template('stats.html', **locals())