segfault in gendb for high-order markov models

[Tomás Pires de Carvalho Gomes]

Hi all,

When using gendb (from MEME version 4.11.2) I will get a segfault when trying to use an order greater than 0, when a background is supplied. The output is as follows:

gendb -type 1 -order 1 -seed 1 -bfile background_example.txt 1

>SEQ_1 1

Segmentation fault

The background file used is the following (although this as never worked with any background):

# 1-order Markov frequencies from file /path/to/fasta

# seqs: 1352    min: 87    max: 2635    avg: 431.8    sum: 583732    alph: DNA

# order 0

A 2.017e-01

C 2.983e-01

G 2.983e-01

T 2.017e-01

# order 1

AA 4.899e-02

AC 4.835e-02

AG 7.364e-02

AT 3.084e-02

CA 6.459e-02

CC 9.456e-02

CG 6.539e-02

CT 7.364e-02

GA 5.992e-02

GC 9.546e-02

GG 9.456e-02

GT 4.835e-02

TA 2.821e-02

TC 5.992e-02

TG 6.459e-02

TT 4.899e-02

Thanks in advance,

Tomás

[CharlesEGrant]

Hi Tomas,

It looks like we introduced a bug into gendb when we added support for arbitrary alphabets. We working on a fix and will have a patch shortly. In the meantime gendb will work as long as you stick to background models of order 0, or omit the background model entirely.

Our apologies for the inconvenience. We'll let you know as soon as we have a fix.

Charles

[Tomás Pires de Carvalho Gomes]

Hi Charles,

Thank you so much! Looking forward to it!

Tomás