Thomas Guiraud

Laboratoire COBIOTEX · Research and development

Hi there,

I would like to share with you a problem that I commonly encounter on NCBI's BLAST web server:

I would like to recover the homologous nucleotide sequences of a gene of interest (presumed single copy) within a microbial taxon (most often within a genus).

As a starting sequence, I use a protein sequence and not a nucleotide sequence: if I use a DNA sequence from one of the species of the genus studied, I risk missing the homologous genes in the most distant species within the genre. So I perform a TBLASTN.

When I perform this TBLASTN on the NCBI BLAST server, I adapt the parameters so as to target the taxon studied, to recover a maximum of sequences (I target 1000 hits or even 5000 hits depending on the situation) and I use a fairly stringent e-value threshold (I do tests before) which ranges from e-50 to e-100 or even more stringent.

As databases, I target either the NR base or the WGS base depending on the situation. I specify it in case I am asked the question but it does not change much to the problem encountered.

The results displayed by the server seem to be suitable most of the time and I have in front of me the list of strains which have been fully sequenced to date and for which a sequence homologous to my sequence of interest has been found by the algorithm.

On the other hand, when I try to download the fasta file of the aligned sequences, I recover a number of sequences much larger than what was displayed. Tests have shown me that this number depends on the stringency chosen in the parameters. For example, if I keep the default e-value settings (0.05), I can potentially recover over 100,000 sequences with sometimes over 100 sequences per accession!

Already, if you have a trick to recover only the most homologous sequences, I'm interested. But my main problem is this:

In the recovered Fasta file, the downloaded sequences are not ordered according to their e-value, their homology but according to their position in the genome (or contig). It would have been possible for me in a few clicks to sort the hits to retrieve only the first of the sequences for each accession, but under these conditions, the sorting is too laborious.

Have you faced this problem and if so, how did you solve it (if you solved it…) ?

Thank you for your attention.

In parallel with a project I am currently working on ("BiodiverCité", the inventory and characterization of the wetlands of the urban area of Bordeaux in France), partners will have to opportunity to collect samples of water over a few dozen streams carefully selected. While they are interested in the hydrological and chemical characteristics of these rivers and their associated fauna, I had the idea to take advantage of this sampling campaign to focus on the microflora and, later, possibily establish a link between the different data produced.

I am interested in two types of microflora : on the one hand, bacteria, archaea and other unicellular organisms (but mainly bacteria I must admit) and, on the other hand, the viral microflora.

Concerning the bacterial flora, I will unfortunately exclude any pasteurian approach, because of lack of time and lack of means.

I started with the idea of centrifuging the collected water samples and carrying out a DNA extraction which would join a DNA library to be analyzed once the financial means have been collected.

For the viral flora, I thought to carry out a precipitation of 1M NaCl, PEG6000 10% on the supernatant to concentrate the viruses in a volume easily handled and storable to have a phagolibrary in which to possibly find phages directed against bacterial species of interest (eg F. psychrophilum or other pathogenic species in aquaculture) and secondly, carry out a phage nucleic acid extraction to constitute a phagoDNAlibrary exploitable in the longer term.

My questions are:

1) What volume of sample should be collected to have enough bacterial DNA to make metabarcoding ? For the soil, the specialists of Genosol (INRA, France) urge me to extract from at least 500 mg of soil for bacterial and 1 g for fungal. But for water ?

2) Is the NaCl / PEG6000 precipitation suitable for a long-term phago-library ? I fear that this step is too brutal for some phages and that even if I concentrate theoretically in number of viral particles, I lose in infectivity (due to damage to viral fibrils). If this technique is suitable, is the use of PEG8000 better recommended to ensure a wider enrichment spectrum (for podoviridae for example, more difficult to precipitate if I understand my literature) ?

3) I know there are publications on viromics but I have not looked into detail: what minimum sample volume (before precipitation) is desirable to have enough DNA for metagenomics (or just qPCR detection of well known virus) ? 200 ml ? 1L ? 10L ??? (2 ml :) ?)

Thank you for your attention.