Strange length of merged paired-end sequencing reads after using BBmerge

Question

[bioplanet]

Hi all,

I encountered a strange problem when using BBmerge to align paired-end sequencing data. My construct that got amplified is 430nts long and each read is of 250nts long. However, out of the 2,2M reads, I get some strange lengths for the merged pairs). As you will see in the list below (where I list lengths <250), the pairs with length of 58, 59 and 60 are ~10,500 and I have no idea why this is even possible. My question is therefore if someone can explain this to me and also if I should discard merged reads that do not have a length of at least 250nt (i.e. equal to one of the 2 reads that are merged).

LENGTH #READ PAIR
60  5753
59  4294
58  544
239 43 
221 37
222 29
215 28
230 27
248 26
213 26
197 26
200 25
204 24
228 23
226 23
219 23
249 22
241 21
247 20
246 20
244 20
232 20
203 20
223 19
216 19
211 19
205 19
231 17
192 17
187 16
182 16
238 15
229 15
207 15
202 15
198 15
176 15
233 14
227 14
224 14
208 14
174 14
236 13
234 13
218 13
206 13
189 13
188 13
243 12
186 12
240 11
212 11
172 11
220 10
214 10
199 10
178 10
177 10
162 10
142 10
140 10
237 9
225 9
194 9
161 9
144 9
210 8
201 8
175 8
169 8
168 8
163 8
148 8
146 8
242 7
193 7
191 7
173 7
171 7
153 7
127 7
114 7
245 6
217 6
196 6
185 6
183 6
164 6
160 6
150 6
135 6
130 6
128 6
116 6
195 5
190 5
184 5
165 5
159 5
157 5
154 5
151 5
134 5
132 5
131 5
125 5
109 5
235 4
209 4
181 4
167 4
155 4
147 4
141 4
124 4
100 4
88  4
179 3
166 3
156 3
139 3
136 3
123 3
122 3
120 3
118 3
117 3
113 3
90  3
170 2
152 2
149 2
138 2
137 2
133 2
129 2
119 2
115 2
107 2
106 2
105 2
103 2
101 2
96  2
87  2
180 1
158 1
145 1
143 1
112 1
110 1
97  1
95  1
91  1
84  1
81  1
80  1
78  1
74  1
67  1
66  1
57  1
45  1
44  1
40  1

[GenoMax]

Was the insert/library size selected? There was no fragmentation done, correct? Are you merging the reads before doing any trimming? Have you scanned/trimmed this data after merging?