varvamp 1.2.2__py3-none-any.whl → 1.3__py3-none-any.whl

varvamp/scripts/default_config.py

@@ -9,7 +9,7 @@ __all__ = [
     'PCR_DNA_CONC', 'PCR_DNTP_CONC', 'PCR_DV_CONC', 'PCR_MV_CONC',
     'PRIMER_3_PENALTY', 'PRIMER_GC_END', 'PRIMER_GC_PENALTY',
     'PRIMER_GC_RANGE', 'PRIMER_HAIRPIN', 'PRIMER_MAX_BASE_PENALTY',
-    'PRIMER_MAX_DIMER_TMP', 'PRIMER_MAX_DINUC_REPEATS', 'PRIMER_MAX_POLYX',
+    'PRIMER_MAX_DIMER_TMP', 'PRIMER_MAX_DIMER_DELTAG', 'PRIMER_MAX_DINUC_REPEATS', 'PRIMER_MAX_POLYX',
     'PRIMER_MIN_3_WITHOUT_AMB', 'PRIMER_PERMUTATION_PENALTY',
     'PRIMER_SIZES', 'PRIMER_SIZE_PENALTY',
     'PRIMER_TMP', 'PRIMER_TM_PENALTY',
@@ -30,7 +30,10 @@ PRIMER_MAX_DINUC_REPEATS = 4  # max number of polyXY
 PRIMER_HAIRPIN = 47  # max melting temp for secondary structure
 PRIMER_GC_END = (1, 3)  # min/max GCs in the last 5 bases of the 3' end
 PRIMER_MIN_3_WITHOUT_AMB = 3  # min len of 3' without ambiguous charaters
-PRIMER_MAX_DIMER_TMP = 47  # max melting temp for dimers (homo- or heterodimers)
+# primer dimer constraints
+PRIMER_MAX_DIMER_TMP = 35  # max melting temp for dimers, lower temperature means more stringent filtering
+PRIMER_MAX_DIMER_DELTAG = -9000  # max allowed gibbs free energy for dimer formation, higher values mean more stringent filtering
+END_OVERLAP = 5  # maximum allowed nt overlap between ends of primers to be considered a dimer
 
 # QPCR parameters
 # basic probe parameters
@@ -42,7 +45,6 @@ QPROBE_GC_END = (0, 4)
 QPRIMER_DIFF = 2  # maximal temperature diff of qPCR primers
 QPROBE_TEMP_DIFF = (5, 10)  # min/max temp diff between probe and primers
 QPROBE_DISTANCE = (4, 15)  # min/max distance to the primer on the same strand
-END_OVERLAP = 5  # maximum allowed nt overlap between the ends of probe and primer
 QAMPLICON_LENGTH = (70, 200)  # min/max length of the qPCR amplicon
 QAMPLICON_GC = (40, 60)  # GC min/max of the qPCR amplicon
 QAMPLICON_DEL_CUTOFF = 4  # consider regions of the alignment for deltaG calculation if they have smaller deletions than cutoff

varvamp/scripts/logging.py

@@ -9,6 +9,7 @@ import shutil
 import datetime
 import random
 import statistics
+import re
 
 # varVAMP
 from varvamp.scripts import config
@@ -106,13 +107,12 @@ def raise_arg_errors(args, log_file):
             "degeneracy. Consider reducing.",
             log_file
         )
-    if args.database is not None:
-        if args.threads < 1:
-            raise_error(
-                "number of threads cannot be smaller than 1.",
-                log_file,
-                exit=True
-            )
+    if args.threads < 1:
+        raise_error(
+            "number of threads cannot be smaller than 1.",
+            log_file,
+            exit=True
+        )
     if args.mode in ("tiled", "single"):
         if args.opt_length > args.max_length:
             raise_error(
@@ -152,9 +152,9 @@ def raise_arg_errors(args, log_file):
                 "small overlaps might hinder downstream analyses. Consider increasing.",
                 log_file
             )
-        if args.overlap > args.max_length / 2 - config.PRIMER_SIZES[1]:
+        if args.overlap > args.max_length / 2 - config.PRIMER_SIZES[0] * 2:
             raise_error(
-                "min overlap must be lower than half of your maximum length - maximum primer length. To achieve optimal results reduce it to at least half of your optimal length",
+                "min overlap must be lower than your maximum length / 2 - 2 * min primer length.",
                 log_file,
                 exit=True
             )
@@ -164,9 +164,9 @@ def raise_arg_errors(args, log_file):
                 log_file,
                 exit=True
             )
-        if args.overlap > args.opt_length / 2:
+        if args.overlap > args.opt_length / 2 - config.PRIMER_SIZES[0] * 2:
             raise_error(
-                "your intended overlap is higher than half of your optimal length. This reduces how well varvamps will find overlapping amplicons. Consider decreasing.",
+                "your min overlap is lower than your optimal length / 2 - 2 * min primer length. This reduces how well varvamps will find overlapping amplicons. Consider decreasing.",
                 log_file
             )
     # QPCR specific warnings
@@ -243,6 +243,34 @@ def check_alignment_length(preprocessed_alignment, log_file):
         )
 
 
+def check_gaped_sequences(preprocessed_alignment, log_file):
+    """
+    checks the number of gaps in each sequence of the alignment and reports a warning
+    if the number of gaps is larger than the mean + 3std
+    """
+    number_of_gaps = {}
+
+    for seq in preprocessed_alignment:
+        # find all gaps for all sequences with regular expression -{min}
+        number_of_gaps[seq[0]] = len(list(re.finditer(r"-+", seq[1])))
+
+    # clac mean and std
+    mean_gaps, mean_gaps_std = statistics.mean(number_of_gaps.values()), statistics.stdev(number_of_gaps.values())
+
+    warning = []
+
+    for name, n_gaps in number_of_gaps.items():
+        if n_gaps < mean_gaps - 3 * mean_gaps_std:
+            warning.append(f"{name} ({n_gaps} gaps)\n")
+
+    if warning:
+        raise_error(
+            f"The following sequences contain less gaps than the alignment mean ({round(mean_gaps)} gaps) and might overproportionally gap the alignment:\n{"".join(warning)}",
+            log_file,
+            exit=False
+        )
+
+
 def confirm_config(args, log_file):
     """
     checks the config. raises error and warnings
@@ -252,7 +280,7 @@ def confirm_config(args, log_file):
 
     # check if all variables exists
     all_vars = [
-        # arg independent TILED, SINGLE mode
+        # arg independent all modes
         (
             "PRIMER_TMP",
             "PRIMER_GC_RANGE",
@@ -262,6 +290,7 @@ def confirm_config(args, log_file):
             "PRIMER_MAX_DINUC_REPEATS",
             "PRIMER_GC_END",
             "PRIMER_MAX_DIMER_TMP",
+            "PRIMER_MAX_DIMER_DELTAG",
             "PRIMER_MIN_3_WITHOUT_AMB",
             "PCR_MV_CONC",
             "PCR_DV_CONC",
@@ -272,7 +301,8 @@ def confirm_config(args, log_file):
             "PRIMER_SIZE_PENALTY",
             "PRIMER_MAX_BASE_PENALTY",
             "PRIMER_3_PENALTY",
-            "PRIMER_PERMUTATION_PENALTY"
+            "PRIMER_PERMUTATION_PENALTY",
+            "END_OVERLAP"
         ),
         # arg independent QPCR mode
         (
@@ -283,7 +313,6 @@ def confirm_config(args, log_file):
             "QPRIMER_DIFF",
             "QPROBE_TEMP_DIFF",
             "QPROBE_DISTANCE",
-            "END_OVERLAP",
             "QAMPLICON_LENGTH",
             "QAMPLICON_GC",
             "QAMPLICON_DEL_CUTOFF"
@@ -379,7 +408,7 @@ def confirm_config(args, log_file):
         ("max base penalty", config.PRIMER_MAX_BASE_PENALTY),
         ("primer permutation penalty", config.PRIMER_PERMUTATION_PENALTY),
         ("qpcr flanking primer difference", config.QPRIMER_DIFF),
-        ("probe and primer end overlap", config.END_OVERLAP),
+        ("end overlap", config.END_OVERLAP),
         ("qpcr deletion size still considered for deltaG calculation", config.QAMPLICON_DEL_CUTOFF),
         ("maximum difference between primer and blast db", config.BLAST_MAX_DIFF),
         ("multiplier of the maximum length for non-specific amplicons", config.BLAST_SIZE_MULTI),
@@ -417,15 +446,20 @@ def confirm_config(args, log_file):
         )
     if config.PRIMER_HAIRPIN < 0:
         raise_error(
-            "decreasing hairpin melting temp to negative values "
-            "will influence successful primer search!",
+            "decreasing hairpin melting temp to negative values will influence successful primer search!",
+            log_file
+        )
+    if config.PRIMER_MAX_DIMER_TMP < 21:
+        raise_error(
+            "there is no need to set max dimer melting temp below room temperature.",
             log_file
         )
-    if config.PRIMER_MAX_DIMER_TMP < 0:
+    if config.PRIMER_MAX_DIMER_DELTAG > -6000:
         raise_error(
-            "there is no need to set max dimer melting temp below 0.",
+            "primer interactions with deltaG values higher than -6000 are generally considered unproblematic.",
             log_file
         )
+
     if config.PRIMER_MAX_BASE_PENALTY < 8:
         raise_error(
             "decreasing the base penalty will filter out more primers.",
@@ -566,7 +600,6 @@ def goodbye_message():
     messages = [
         "Thank you. Come again.",
         ">Placeholder for your advertisement<",
-        "Make primers great again!",
         "Ciao cacao!",
         "And now lets pray to the PCR gods.",
         "**bibobibobop** task finished",
@@ -582,6 +615,19 @@ def goodbye_message():
         "Task failed successfully.",
         "Never gonna give you up, never gonna let you down.",
         "Have you tried turning it off and on again?",
+        "Just try it. PCR is magic!",
+        "One goat was sacrificed for this primer design to work.",
+        "You seem trustworthy. Here's a cookie!",
+        "Owwww yeah, primers done!",
+        "These primers were designed without AI assistance.",
+        "Research is fun (if you ignore the pipetting).",
+        "Balance your primers, balance your life.",
+        "Keep calm and PCR on.",
+        "In silico we trust.",
+        "May the primers be with you.",
+        "Designing primers like a boss!",
+        "Primer design completed. Time for a break!",
+        "Eureka! Your primers are ready.",
         "Look, I am your primer scheme.",
         "Quod erat demonstrandum.",
         "Miau?",

varvamp/scripts/param_estimation.py

@@ -1,36 +1,38 @@
 """
-estimate varVAMP threshold and max n of ambiguous bases if none are given
+estimate varVAMP threshold if not given
 """
 
+# libs
+import numpy as np
+
 # varVAMP
 from varvamp.scripts import config
 
 
 def calculate_frequencies(preprocessed_alignment):
     """
-    calculate the max nucleotide freq at each pos
+    calculate individual frequencies for a, t, c, g, and gaps at each position
+    returns a 2D numpy array where each column is a position and rows are [a, c, t, g, -]
     """
-    all_freqs = []
+    # convert alignment to numpy array
+    alignment_array = np.array([list(seq.lower()) for _, seq in preprocessed_alignment])
+    num_sequences, sequence_length = alignment_array.shape
+
+    # calculate frequencies
+    frequencies = np.zeros((5, sequence_length))
+    nucleotides = ['a', 'c', 't', 'g', '-']
 
-    for i in range(0, len(preprocessed_alignment[0][1])):
-        nuc_dict = {
-            "a": 0,
-            "c": 0,
-            "t": 0,
-            "g": 0,
-        }
-        for seq in preprocessed_alignment:
-            if seq[1][i] in nuc_dict:
-                nuc_dict[seq[1][i]] += 1
-        highest_freq = max(nuc_dict.values())
-        all_freqs.append(highest_freq/len(preprocessed_alignment))
+    for i in range(sequence_length):
+        column = alignment_array[:, i]
+        for nuc_idx, nuc in enumerate(nucleotides):
+            frequencies[nuc_idx, i] = np.count_nonzero(column == nuc) / num_sequences
 
-    return all_freqs
+    return np.max(frequencies[0:4, :], axis=0), frequencies[4, :]
 
 
-def calculate_distances(all_freqs, threshold):
+def calculate_distances(highest_freq, deletion_freq, threshold):
     """
-    calc the distance for each  nuc freq to the prior
+    calc the distance for each nuc freq to the prior
     nuc freq that fell below the cutoff
     """
 
@@ -38,11 +40,14 @@ def calculate_distances(all_freqs, threshold):
     previous_dis = 0
     all_dis = []
 
-    for idx, freq in enumerate(all_freqs):
+    for idx, (freq, del_freq) in enumerate(zip(highest_freq, deletion_freq)):
         if freq < threshold:
             current_dis = 0
+        # ignore gaps -> they will be excluded later and might skew distance calc
+        if del_freq > 1-threshold:
+            continue
         current_dis += 1
-        if current_dis <= previous_dis or idx == len(all_freqs)-1:
+        if current_dis <= previous_dis or idx == len(highest_freq)-1:
             all_dis.append(previous_dis)
 
         previous_dis = current_dis
@@ -54,53 +59,70 @@ def get_parameters(preprocessed_alignment, args, log_file):
     """
     give an estimate for number of ambiguous chars and/or threshold
     writes to log file
+
+    How it works:
+
+    - calculate a np array with frequencies per position for ATCG and gaps (-)
+    - increment for different threshold starting at 0.1
+    - for each increment calculate:
+        - the lengths of conseq. nucleotides that are >= threshold
+        - skip all positions with gaps >= 1 - threshold
+        - this results in an array of stretch lengths
+        - Then check for each stretch if it could be considered as a primer region:
+            e.g. 4 ambiguous bases:
+                stretches: [1,1,1,1,6,10,5,3,1,1]
+                - first check for the current stretch if the sum of the next n stretches (with n being
+                the n ambiguous bases + 1) is larger than the minimal length of a primer
+                - this would result that the with stars marked positions would pass:
+                [1,1,1,1*,6*,10*,5,3,1,1]
+        - now calculate the distance between the current start of a stretch and the previous stop
+        - if this is larger than the minimal amplicon length - exit the optimization
+        - reset the threshold to second prior iteration to make it more robust (allow more primer regions) as
+        sometimes the optimization would fail with just one iteration
+
+    Manual optimization can be beneficial in some cases.
     """
     # set coverage to max
-    coverage = 1
-    # read in the alignment and calc freqs
-    frequencies = calculate_frequencies(preprocessed_alignment)
-    # if no args for both threshold and n_ambig are given
-    # set the n_ambig to 2 and optimize threshold
-    if args.threshold is None:
-        args.threshold = 0.1
-        if args.n_ambig is None:
-            args.n_ambig = 2
-        text = f"AUTOMATIC PARAMETER SELECTION\nvarVAMP estimates the threshold at {args.n_ambig} ambiguous bases"
-        fixed = False
-    # if threshold is given, optimize n_ambig (number of ambiguous chars)
-    else:
-        args.n_ambig = config.PRIMER_SIZES[0]
-        text = f"varVAMP estimates the number of ambiguous bases at a threshold of {args.threshold}"
-        fixed = True
+    args.threshold = 0.1
+    # calc freqs
+    highest_frequencies, deletion_frequency = calculate_frequencies(preprocessed_alignment)
     # write to log
     with open(log_file, 'a') as f:
-        print(f"{text}\nto consider ~50% of the alignment for potential primers:\n\n-t\t-a\testimated coverage", file=f)
-
-        # optimize until less than 50 % is covered
-        while coverage >= 0.5 and args.threshold < 1:
-            distances = calculate_distances(frequencies, args.threshold)
-            # calculate the cummulative sum of the sum of n conseq. streches
-            # that are together larger than the min primer length
-            covered_pos = sum(
-                [distances[x] for x in range(0, len(distances)) if sum(distances[x:x+args.n_ambig+1]) >= config.PRIMER_SIZES[0]]
-            )
-            # calculate coverage
-            coverage = (covered_pos+1)/len(preprocessed_alignment[0][1])
-            # change the non fixed param if threshold has not been reached
-            if coverage >= 0.5:
-                # write each iteration to log
-                print(round(args.threshold, 2), args.n_ambig, round(coverage*100, 1), sep="\t", file=f)
-                if fixed:
-                    args.n_ambig -= 1
-                else:
+        print(f"AUTOMATIC THRESHOLD SELECTION\n", file=f)
+        print(f"-t\tmaximum non-covered region", file=f)
+        # calculate distance between passing potential primer regions
+        previous_stop = 0
+        while args.threshold < 1:
+            distances = calculate_distances(highest_frequencies, deletion_frequency, args.threshold)
+            max_distance_between_passing, previous_passing = 0, 0
+            # check if the distance between potential primer regions is not larger than:
+            # args.opt_length - 2 * args.overlap
+            for idx, dis in enumerate(distances):
+                if sum(distances[idx:idx + 1 + args.n_ambig]) >= config.PRIMER_SIZES[1]:
+                    # the stretch start in the gap-excluded alignment is the sum of all prior distances including the current
+                    # minus the distance of the current stretch
+                    stretch_start = sum(distances[:idx])
+                    # then the distance between the prior stop and current start is calculated
+                    current_dis = stretch_start - previous_stop
+                    # and the max is updated if necessary
+                    if max_distance_between_passing < current_dis:
+                        max_distance_between_passing = current_dis
+                    # update previous stop position
+                    previous_stop = stretch_start + distances[idx]
+            # write each iteration to log
+            print(round(args.threshold, 2), max_distance_between_passing, sep="\t", file=f)
+            # check if the distance is acceptable
+            distance_threshold = args.opt_length - 2 * args.overlap if args.mode == 'tiled' else args.opt_length
+            if max_distance_between_passing < distance_threshold:
+                # never exceed 0.99
+                if args.threshold < 0.99:
                     args.threshold += 0.01
-            # or reset to the param of the prior iteration
-            else:
-                if fixed:
-                    args.n_ambig += 1
                 else:
-                    args.threshold -= 0.01
+                    break
+            # or reset to the param of the two previous iterations
+            else:
+                args.threshold -= 0.02
                 break
-        print(f"Automatic parameter selection set -t {round(args.threshold, 2)} and -a {args.n_ambig}.", file=f)
+        print(f"Automatic parameter selection set -t {round(args.threshold, 2)} at -a {args.n_ambig}.", file=f)
 
-    return round(args.threshold, 2), args.n_ambig
+    return round(args.threshold, 2)